standing on the shoulders of giants: the effect of passive ... · [14:32 19/10/2018...

[14:32 19/10/2018 RFS-OP-REVF180110.tex] Page: 1 1–55

Standing on the Shoulders of Giants: TheEffect of Passive Investors on ActivismIan R. AppelCarroll School of Management, Boston College

Todd A. GormleyOlin Business School, Washington University in St. Louis

Donald B. KeimThe Wharton School, University of Pennsylvania

We analyze whether the growing importance of passive investors has influenced thecampaigns, tactics, and successes of activists. We find activists are more likely to seekboard representation when a larger share of the target company’s stock is held bypassively managed mutual funds. Furthermore, higher passive ownership is associated withincreased use of proxy fights, settlements, and a higher likelihood the activist achievesboard representation or the sale of the targeted company. Our findings suggest that therecent growth of passive institutional investors mitigates free-rider problems and facilitatesactivists’ ability to engage in costly, value-enhancing forms of monitoring. (JEL D22,G23, G30, G34)

Received September 28, 2016; editorial decision August 18, 2018 by Editor AndrewKarolyi.

We thank Nicole Boysen, Alon Brav, Alex Edmans, Rüdiger Fahlenbrach, Slava Fos, Nickolay Gantchev,Christopher Hennessey, Travis Johnson, Andrew Karoyli, Hyunseob Kim, Gilberto Loureiro, Nadya Malenko,Pedro Matos, Jordan Nickerson, Min Pyo, Christopher Schwarz, and Jules van Binsbergen; seminar and brown bagparticipants at Baylor University (Hankamer), Boston College (Carroll), Cornell University (Dyson), ErasmusUniversity (School of Economics), Georgia Tech (Scheller), Lancaster University, London Business School,London School of Economics, Maastricht University, Pennsylvania State University (Smeal), Tilburg University,University of Alabama (Culverhouse), University of Amsterdam, University of Arizona (Eller), Universityof Massachusetts Boston (College of Management), University of Pennsylvania (Wharton), and WashingtonUniversity in St. Louis (Olin); and participants at the Carnegie Mellon University Accounting Conference,European Finance Association Annual Meeting (Norway), 2017 Financial Intermediation Research SocietyConference, FTSE 2016 World Investment Forum, 2016 FSU SunTrust Beach Conference, 2016 NationalBureau of Economic Research Conference on Long-Term Asset Management, 2016 Nova SBE – BancoBPI Corporate Finance Conference, 2016 Norwegian Financial Research Conference, 2017 Swedish Houseof Finance Conference on Institutional Investors and Corporate Governance, 2016 University of KentuckyFinance Conference, and 2016 Western Finance Association Annual Meeting for helpful comments. We thankLouis Yang for his research assistance, Alon Brav for sharing activism data, and the Rodney L. White Centerfor Financial Research for financial support. Send correspondence to Todd Gormley, Washington Universityin St. Louis, One Brookings Drive, Campus Box 1133, St. Louis, MO 63130; telephone: (314) 935-7171.E-mail: [email protected].

© The Author(s) 2018. Published by Oxford University Press on behalf of The Society for Financial Studies.All rights reserved. For permissions, please e-mail: [email protected]:10.1093/rfs/hhy106


The Review of Financial Studies / v 00 n 0 2018

The willingness of investors to engage in activism has rapidly grown in recentyears. Hundreds of activist campaigns are launched annually, and, as noted byThe Economist (2015, p. 11), the current “scale of their insurrection in Americais unprecedented.” Evidence also suggests the goals of activists have becomemore ambitious, and their success rate has improved. For example, activistsincreasingly wage proxy fights to obtain board representation, and more than70% of these campaigns were successful in 2014.1

At the same time, stock ownership by passive institutional investors hasgrown rapidly. Passively managed mutual funds, which seek to deliver thereturns of a market index (e.g., S&P 500) or particular investment style (e.g.,large-cap value), have quadrupled their ownership share of the U.S. stockmarket over the last 15 years and now account for more than one-third ofall mutual fund assets. Institutions that offer these funds, like Vanguard andBlackrock, are often the largest shareholders of U.S. companies. In this paper,we examine whether these two concurrent trends are related. In particular, weanalyze whether the increasingly large and concentrated ownership stakes ofpassive institutional investors influence the types of campaigns undertaken byactivists, the tactics they employ, and their eventual outcomes.

One possibility is that the increased presence of passive institutions facilitatesactivism. Activist investors face a classic free-rider problem (Grossman andHart, 1980) when considering intervention in a firm; that is, the activistbears all costs associated with intervention, yet the benefits accrue acrossall shareholders. The large and concentrated ownership stakes of passiveinstitutions might help overcome this problem by facilitating activist investors’ability to rally support for their demands (Brav et al. 2008; Bradley et al.2010) and by decreasing the coordination costs of activism (e.g., during theproxy solicitation process). The inability of passive institutions to sell poorlyperforming stocks in their portfolios (because of their mandate to closely trackunderlying indexes) might also make them more willing and influential partnersin an activist campaign than other shareholders, and their support may lendcredibility to campaigns and ultimately increase the likelihood of a successfuloutcome.

However, it is also possible that the growing clout of passive institutionsmight hamper activism. If passive investors “take little interest in how firms arerun …[and] dislike becoming deeply involved in management” (The Economist2015, p. 11), their increasing market share could make it more difficult foractivists to rally support for their demands. Some activists also argue that passiveinstitutions have a conflict of interest. In particular, a fear of losing the businessof corporate pension plans, one of the largest investors in index funds, may deter

1 For example, Hoffman and Benoit (2015) report that the number of companies targeted by an activist seekingboard representation has more than doubled in the last 5 years. Benoit (2014b) reports that activists seeking aboard seat obtained at least a partial victory in 72% of such campaigns in 2014, up from a success rate of 57%in 2008.

2


Standing on the Shoulders of Giants

such institutions from supporting activists.2 Finally, as long-term investors,passive institutions might not always share the same goals as activists. Forexample, both Blackrock and State Street, two of the largest institutions offeringindex funds, have expressed unwillingness to support activist demands they seeas shortsighted, including demands for increased debt and payouts.3

Identifying the impact of passive investors on activists’ choices and successrates poses an empirical challenge. The primary concern is that of omittedvariable; because passive institutional portfolios are related to the compositionof the indexes they track, passive ownership of a stock might be correlated withfactors that directly affect activists’ tactics and success rates. For example, poorpast performance might cause both a stock’s removal from a popular index,thus reducing passive ownership, and also increase the likelihood of activism.Thus, naïve correlations between passive institutional ownership and activismoutcomes might not reflect a causal relation.

To overcome this challenge, we exploit variation in stock ownership bypassive mutual funds that occurs around the cutoff point used to construct twowidely used market benchmarks: the Russell 1000 and Russell 2000 indexes. Asshown in Appel, Gormley, and Keim (2016) (hereafter AGK) among others,benchmarking by passive funds leads to a sharp difference in ownership bypassive investors for stocks at the top of the Russell 2000 relative to stocksat the bottom of the Russell 1000 even though they are otherwise similar. Forexample, during our sample period from 2008 to 2014, ownership by passivelymanaged mutual funds and exchange-traded funds (ETFs) is about 40% higher,on average, for stocks at the top of the Russell 2000 index relative to otherwisesimilar stocks at the bottom of the Russell 1000 index.

Previous work has exploited the Russell 1000/2000 cutoff to analyzepassive investors’ ability to directly influence firms’ governance structures(e.g., board composition and voting rights), but we use this setting to analyzewhether passive investors help or hinder activists, an entirely separate class ofinstitutional investors widely thought to play an important role in monitoringmanagers. Specifically, we assess passive investors’ impact on the campaigns,tactics, and successes of activists by following the approach of AGK andinstrumenting ownership by passive funds with an indicator for assignment tothe Russell 2000. However, because our sample of activism events runs through2014, which is necessary to study the recent shifts in activism outcomes, weaugment the specification of AGK to account for an important change in howRussell constructed the two indexes after 2006 (see Section 2.1 for details).Our IV estimation relies on the assumption that, after conditioning on factorsthat determine assignment to either the Russell 1000 or Russell 2000 index,

2 For example, see the annual letter of hedge fund manager William A. Ackman (2016) to the investors of PershingSquare Capital Management in January 2016. See also Henderson and Shapiro Lund (2017).

3 See Benoit (2014a) and Kumar and O’Hanley (2016).

3



inclusion in the Russell 2000 index does not directly affect activist outcomesexcept through its impact on passive ownership.

Using our IV approach, we find that passive mutual funds significantly affectactivists’ strategic choices. While the percentage of a firm’s stock held bypassive mutual funds is not associated with the likelihood of being targetedby an activist, we show that, conditional on being targeted by an activist,the percentage of stock held by passive investors is significantly related toactivists’ campaign goals. Specifically, among firms targeted by an activist,a 1-standard-deviation increase in passive ownership, which is approximately3.6 percentage points relative to an average of 9.4% for our sample of activistevents, is associated with about a 30- to 36-percentage-point increase in theproportion of campaigns seeking board representation, and a similar magnitudedecrease in other types of campaigns, including those limited to shareholderproposals and exempt solicitations. The increase in campaigns seeking boardrepresentation is economically large, corresponding to a doubling in its overallfrequency, and suggests activists set relatively more ambitious goals when moreof a company’s stock is held by passive investors.

We also find that greater passive ownership is associated with the increaseduse of confrontational tactics by activists. While board representation canbe gained through both friendly and confrontational approaches (Brav et al.2008; Fos 2017), we document a shift in the likelihood of activists employinghostile tactics in attempts to gain board seats when passive ownership ishigher. Specifically, among firms targeted by an activist, a 1-standard-deviationincrease in passive ownership is associated with over a 150% increase in thelikelihood of activists launching a proxy fight against incumbent directors.Furthermore, we find an increase in the total number of board seats soughtwhen passive ownership is higher; a 1-standard-deviation increase in passiveownership is associated with one additional board seat being sought by theactivist, relative to an average of 0.76 seats sought.

Combined, our results suggest that the presence of passive institutionsincreases activists’ willingness to engage in costlier forms of activism. The costsassociated with seeking board representation and initiating a proxy fight canamount to millions of dollars (Gantchev 2013), while pushing for a shareholderproposal or exempt solicitation is “easier, less costly and demand a lower level ofcommitment from dissidents” (Wilcox 2005). Consistent with this shift towardmore costly forms of activism, we also find that activists are more likely to seekreimbursement from the company for their campaign when passive ownershipis higher. The increased willingness to undertake such campaigns could reflectlower expected costs of such campaigns (e.g., lower coordination costs) and/orhigher expected benefits (e.g., increased likelihood of winning) when a largerproportion of a firm’s equity is held by passive investors.

Consistent with passive funds facilitating activists’ ability to engage in costlyforms of monitoring, higher passive ownership also increases activists’ successrates in achieving changes in governance or control. When passive ownership is

4



higher, we document an increase in both the likelihood and favorability of proxysettlements with management. Specifically, a 1-standard-deviation increase inpassive ownership is associated with a 16- to 20-percentage-point increase inthe likelihood of a proxy fight settlement and a 0.27- to 0.35-seat increase innumber of board seats won when settlements occur. The effects are sizable;on average, only 7% of campaigns end with such settlements, and 0.11 boardseats are won during settlements. We also find a positive association betweenpassive ownership and the likelihood of success for campaigns pertaining tocorporate control, including the removal of takeover defenses and the sale ofthe firm to the activist or a third party. In contrast, we do not find more activistsuccesses for outcomes passive investors often associate with short termism,such as increased payouts and changes to the capital structure.

Finally, we find suggestive, but weaker, evidence that higher passiveownership makes it more likely an activism campaign is associated with higherfirm value. Consistent with a positive effect on activists’ ability to seek value-enhancing changes, passive ownership is postively associated with abnormalstock returns around activists’ engagements. A 1-standard-deviation increasein passive ownership is associated with about an average 11-percentage-pointincrease in abnormal stock return at the time of intervention announcement.Our findings regarding announcentment returns, however, are not as robust asour other findings, and selection issues prevent us from directly analyzing anypotential impact on long-term performance. Nevertheless, we find increasesin (1) successful sales of the target firm, which have large positive valueimplications for target shareholders; and (2) settlements, which is wherecampaign-related increases in long-term performance and favorable delistingstend to be concentrated (Bebchuk et al. 2017). Both are consistent withimprovements in long-term value.

Our findings are not driven by a change in the type of firms targeted byactivists and are robust to various specification choices. Specifically, passiveownership is not associated with firm characteristics that have been identifiedin prior research to be related to the likelihood of being targeted by an activist.The findings are also robust to varying the functional form we use to controlfor firms’ end-of-May market cap, to modifying how we measure passive stockownership, to varying the number of firms around the R1000/2000 cutoff thatwe include in our sample, and to adding various controls, including the liquidityof a firm’s stock and whether the firm recently switched indexes. Our findingsare also not sensitive to excluding activists that file a 13D with no stated intentor to only using end-of-May market cap rankings to select our sample of stockseach year. In addition, we find no effect of passive ownership in placebo testsat other, non-Russell 1000/2000 market cap thresholds, providing additionalevidence that our findings are not driven by specification error. Finally, we findsimilar results during our sample period when we use the recently extendedalternative activism data of Brav et al. (2008) and Brav, Jiang, and Kim (2010).Interestingly, we find no evidence of a relation between passive ownership and

5



activism in the earlier years covered by this alternative database, which bothconfirms the growing importance of passive investors for the governance offirms and aligns with anecdotal evidence of a recent shift in activists’ tactics toaccount for passive investors’ increasing influence.4

Overall, this paper contributes to the literature that studies investor activism.A rich literature studies the characteristics of activists’ targets, and theimplications of activism for shareholder value and other corporate outcomeshave been widely studied.5 However, relatively little is understood about howsuch investors choose their tactics and what factors contribute to their success.We contribute to this literature by showing that firms’ ownership structures playa key role in determining the choice of tactics by activists and, ultimately, theoutcomes of their campaigns. Our finding that activists tailor their campaignsto differences in firms’ ownership structures is intuitive and consistent withactivists’ recent shift in tactics and increased successes being at least partlydriven by the growing influence of passive institutional investors. Morover,our finding that passive ownership matters for activist choices primarily at theintensive margin suggests that determinants of being targeted by an activistdiffer from the determinants of activists’ strategic choices once a target isidentified.

Our findings are also related to the recent strand of literature that explorescoordinated actions by “wolf packs” consisting of multiple activists (e.g., Brav,Dasgupta, and Mathews 2015; Coffee and Palia 2016; Dimson, Karakas, andLi 2015). We contribute to this nascent literature by showing that activists’strategic choices also may be influenced by potential alliances with large passiveinstitutional block holders, which represent an increasingly large componentof U.S. stock ownership.6

Finally, we contribute to the growing literature on the governance effects ofpassive institutional investors. For example, AGK find that passive investors usetheir significant voting power to exert direct influence over firms’ governancestructures, whereas Schmidt and Fahlenbrach (2017) argue that passiveinvestors are less effective at engaging in more costly forms of monitoring,such as determining the value of a proposed merger. In contrast to this earlierwork, this paper offers novel evidence that an increased presence of passive

4 In private conversations with the authors, some large passive institutions maintain that their view toward activismhas not changed over the last 10 years, but rather that activists have evolved to be more civil and better focusedon issues of concern to long-term shareholders. This shift in tactics also might have increased the openness ofsome institutions to activists’ demands. See Section 5.2 for more details.

5 See, for example, Brav et al. (2008, 2018), Clifford (2008), Becht et al. (2009), Brav, Jiang, and Kim (2010,2015b), Greenwood and Schor (2009), Klein and Zur (2009), Popadak (2014), Fos and Tsoutsoura (2014),McCahery, Sautner, and Starks (2016), Bebchuk, Brav, and Jiang (2015), Boyson, Gantchev, and Shivdasani(2017), Collin-Dufresne and Fos (2015), and Boyson and Pichler (2018). For comprehensive reviews of thisliterature, see Gillan and Starks (2007), Brav, Jiang, and Kim (2010), and Denes, Karpoff, and McWilliams(2015).

6 See also Kedia, Starks, and Wang (2017) on the importance of “activism-friendly” institutions for activists’successes.

6



investors also affects the choices of activists, a separate class of institutionalinvestors that is known to undertake more costly forms of engagement, like aproxy fight or takeover. Thus, our evidence indicates that, while not engagingin traditional forms of activism themselves or potentially being less effectiveat high-cost monitoring activities, passive investors meaningfully affect theactivism of other investors that specialize in such costly forms of engagement,providing another distinct mechanism by which the recent growth of passiveinvestors is affecting the monitoring of managers.

1. Sample, Data Sources, and Descriptive Statistics

1.1 Mutual fund holdings and Russell 1000/2000 index membershipWe use the S12 mutual fund holdings data compiled by Thomson Reuters andavailable from Wharton Research Data Services (WRDS) to compute mutualfund holdings in a stock as a percentage of its market capitalization. SinceMay 2004, all (open-end) mutual funds and ETFs holding stocks traded onU.S. exchanges are required to report those holdings every quarter to the SECusing Forms N-CSR and N-Q.7 Reported securities include all NYSE, Amex,Nasdaq, Toronto, and Montreal common stocks. We calculate the total marketcap of each stock using the CRSP monthly file as the sum of shares outstandingmultiplied by price for each class of common stock associated with a firm.

To classify a mutual fund as either passively or actively managed, we usethe method of AGK, who flag a fund as passively managed if its fund nameincludes a string that identifies it as an index fund or if the CRSP Mutual FundDatabase classifies the fund as an index fund. We classify all other mutual fundsthat can be matched to the CRSP mutual fund data as actively managed, andfunds that cannot be matched are left unclassified. To generate variables formutual fund ownership disaggregated into these three categories, we computethe percentage of each stock’s market capitalization that is owned by passive,active, and unclassified mutual funds at the end of each quarter. Our calculationconfirms the rise of passive ownership in recent years; Figure 1 illustrates thisby showing a 250% increase in the percentage of equity mutual fund assetsthat are passively managed from 1998 to 2014. Over the same time period,the percentage of total market capitalization that is held by passively managedfunds quadrupled from 1.6% to 8.1%.

Our subsequent analysis is restricted to the sample of stocks in the Russell1000 and 2000 indexes, beginning with the 2007 reconstitution. We start thesample in 2007 to correspond with Russell’s “banding” policy (see Section 2.1for further details) and the availability of our main activism outcomes. RussellInvestments provides index constituents as well as its proprietary measure for

7 Hereafter, we collectively refer to the open-end ETFs in our sample as mutual funds. Closed-end funds, whichare typically actively managed, are excluded from our sample.

7



Figure 1Growth of passive investors, 1998–2014This figure plots the estimated percentage of all U.S. equity mutual fund assets under management that are heldin passively managed funds between 1998 and 2014. We construct the figure by matching the S12 mutual fundholdings data, compiled in the Thomson Reuters Mutual Fund Holdings Database, to fund names in the CRSPmutual fund data. We use a name-parsing procedure along with the index fund identifier from the CRSP mutualfund file to classify mutual funds as passively managed. Our procedure is described in Section 1.1. Holdings andmarket cap are calculated each year at the end of the third quarter.

the float-adjusted market capitalization, which is used to determine the rank(i.e., portfolio weight) of each security within an index.

1.2 Activism dataWe obtain data on corporate activist campaigns from SharkWatch (FactSet),which offers a comprehensive database of activism events after 2005. Thesource of the information in SharkWatch includes company/activist filings andpress releases, news/trade publications, and company Web sites.

We classify activist campaigns into four mutually exclusive categoriesbased on their primary goal: (1) campaigns seeking board representation; (2)campaigns seeking to maximize shareholder value by advocating for specificpolicy changes; (3) all other campaign goals; and (4) 13D filings with no explicitactivist intent. Campaigns seeking board representation capture cases where theactivist attempts to replace either a subset of directors or to take control of theboard. Campaigns seeking change in corporate policies include those where theactivist does not seek board representation but does push for changes thoughtto improve shareholder value, including increased payouts, changes in thecompany’s capital structure, or the sale of the company. Finally, “other goals”generally include less costly campaigns where the activist only seeks an exemptsolicitation, which involve activists communicating with other shareholders, butnot soliciting proxies, or more modest goals like the adoption of a shareholderproposal, defeating a management proposal, removing an officer, and enhancinggovernance structures.

SharkWatch also includes 13D filings with no stated activist goals from 50well-known activists (known as the SharkWatch50). A schedule 13D filing isrequired under Section 12 of the Securities Exchange Act when a shareholder’s

8



beneficial ownership exceeds 5% and that shareholder plans to engage inactivism. The purpose of the transaction (e.g., board representation) also mustbe provided in Item 4 of the 13D filing. Some institutions, however, will file a13D, but not declare specific intent to engage in activism. This is likely doneto leave open the option of becoming more active in the future, and we classifythese campaigns as “13D only.” As discussed in Section 5, however, our mainfindings are robust to excluding activist campaigns associated with 13D filingswith no stated goals.

We also use SharkWatch for data on tactics used by activists and the eventualoutcome of each campaign. Specifically, we construct indicator variables forthe most common tactics employed by activists, including proxy fights, thedrafting of shareholder proposals, or initiating a lawsuit. Finally, we constructindicators for the most common outcomes of an activist campaign: whether theactivist campaign results in a proxy settlement; increased dividends/payouts;governance reform (not including activist representation on the board); oracquisition of the firm by either a third party or the activist.

1.3 Sample and descriptive statisticsFor our main analysis, we restrict the sample to activist events occurring amongthe bottom-500 stocks of the Russell 1000 and the top-500 stocks of the Russell2000 index, as determined using the end-of-June Russell-assigned weights forstocks within each index. There are 466 such events for 310 unique firms, andfor the firms targeted by multiple activist campaigns, 67 are in the same calendaryear. We describe our sampling choice and the inherent trade-offs we face inSection 2.

Table 1 reports summary statistics for our main sample. Total mutual fundownership for the stocks in our sample, is 35.6%. The largest component ofmutual fund ownership is active investors (22.7%), followed by passive (9.4%),and unclassified investors (3.5%). About 28% of all activist campaigns seekboard representation as their primary goal, while seeking to maximize value byenacting policy changes represents 20% of campaigns. The remaining half ofthe campaigns are either classified as an “other campaign type” by SharkWatch(38%) or are campaigns where the investor initiates a 13D filing indicating anintent to engage in activism but does not state a goal or subsequently engagemanagement (14%). Despite their high-profile nature, only about 19% ofcampaigns employ a proxy fight as one of their tactics. About 7% of campaigns(or about 36% of proxy fights) end in a proxy settlement, and activists onlywin proxy fights in 3.2% of campaigns (and 18% of proxy fights) during oursample.

Activists’ success rates can vary considerably across campaign goals. Ofthe 28% of campaigns seeking board representation, activists are successful inattaining this goal in 47.7% of these campaigns. For other desired outcomes,however, activists’ success rates are lower. For example, whereas 13.1% ofcampaigns seek to sell the firm to a third party, which is counted as part of

9



Table 1Summary statistics

Obs. Mean P25 Median P75 SD

Ownership structureTotal mutual fund ownership % 466 35.6 28.2 35.6 43.1 10.8Passive ownership % 466 9.4 6.9 9.1 11.5 3.6Active ownership % 466 22.7 15.3 21.8 29.5 9.7Unclassified ownership % 466 3.5 1.9 2.8 4.7 2.3

Campaign classificationsSeek board representation 466 0.28 0.00 0.00 1.00 0.45Maximize value via policy change 466 0.20 0.00 0.00 0.00 0.40Other 466 0.38 0.00 0.00 1.00 0.4913D filing only 466 0.14 0.00 0.00 0.00 0.35

Proxy fight outcomesProxy fight 466 0.19 0.00 0.00 0.00 0.39Proxy fight - settlement 466 0.07 0.00 0.00 0.00 0.25Proxy fight - activist wins 466 0.03 0.00 0.00 0.00 0.18Proxy fight - firm wins 466 0.03 0.00 0.00 0.00 0.18Proxy fight - withdrawn 466 0.05 0.00 0.00 0.00 0.23Seats sought 466 0.76 0.00 0.00 0.00 1.85# of board seats won in settlement 466 0.11 0.00 0.00 0.00 0.50

Other outcomesAcquisition [by third party] 466 0.02 0.00 0.00 0.00 0.15Acquisition [by activist] 466 0.02 0.00 0.00 0.00 0.13Merger blocked 466 0.04 0.00 0.00 0.00 0.19Removed takeover defense 466 0.04 0.00 0.00 0.00 0.19Increased payouts 466 0.04 0.00 0.00 0.00 0.20Capital structure change 466 0.01 0.00 0.00 0.00 0.10Spinoff 466 0.04 0.00 0.00 0.00 0.19CAR(-10,10) 410 0.05 −0.04 0.03 0.12 0.17

This table reports summary statistics of our key variables for our main sample: activism events that occur forfirms in the 500 bandwidth around the cutoff between the Russell 1000 and 2000 indexes from 2008 to 2014.Table A.1 provides definitions for all variables. We deleted observations where mutual fund ownership is missing.

the 38% of “other campaign types,” reported in Table 1, only 2.4% of thesecampaigns are successful (see Table 1).

2. Empirical Framework

Identifying the impact of passive investors on the types of campaigns undertakenby activists, the tactics they employ, and their eventual outcomes poses anempirical challenge. Cross-sectional correlations between passive ownershipand activism outcomes might not reflect a causal relation because ownership bypassive investors could be correlated with factors, such as firms’ stock liquidityor operating performance, directly affecting activism. Failure to control for suchfactors could introduce an omitted variable bias that confounds inferences. Toovercome this challenge and to determine the importance of passive investors,we use stocks’ assignment to the top of the Russell 2000 index as an exogenousshock to passive mutual fund ownership. We describe our identification strategynext.

10



2.1 Russell index construction and passive institutional investorsPassive funds attempt to match the performance of a market index by holding abasket of representative securities in the particular market index in proportionto their weights in the index. The most visible types of passive funds are indexfunds, which hold nearly all stocks in the market index.

Two market indexes widely used as benchmarks are the Russell 1000 andRussell 2000. During our sample period, the Russell 1000 comprises 1,000U.S. stocks that mostly reflect the largest 1,000 companies in terms of marketcapitalization, while the Russell 2000 comprises the next largest 2,000 stocksnot included in the Russell 1000. To account for changes in stocks’ ranking bymarket cap, the Russell indexes are reconstituted each year at the end of Juneusing a combination of three factors: (1) a stock’s market capitalization as ofthe last trading day in May of that year, (2) the stock’s index assignment inthe previous reconstitution year, and (3) whether the stock’s market cap fallswithin a certain range of the cutoff between 1,000th and 1,001st largest stockmarket caps. Specifically, a stock with an end-of-May market cap below (above)the market cap of the 1,000th (1,001st) largest market cap will be included inthe Russell 2000 (Russell 1000) index unless that stock was included in theRussell 1000 (Russell 2000) last year and its market cap is not below (above)the market cap of the 1,000th (1,001st) largest market cap by more than 2.5%of the cumulative market cap of the Russell 3000E Index, which comprisesthe 4,000 largest stocks. This policy, which Russell refers to as “banding,” wasimplemented in 2007 to minimize the number of stocks that switch indexes eachyear. Prior to 2007, the Russell 1000 simply included the 1,000 largest stocksat the end of the last trading day in May, whereas the Russell 2000 includedthe next 2,000 largest stocks.8

After index assignments are determined, each stock’s weight in the indexis then calculated using its end-of-June float-adjusted market cap. Unlike themarket cap used to determine index membership, the float adjusted market caponly includes the value of shares available to the public. Shares held by anothercompany or individual that exceed 10% of shares outstanding, by anothermember of a Russell index, by an employee stock ownership plan (ESOP),by a government, and those not listed on an exchange are not included whencalculating a firm’s float-adjusted market cap.

Because the Russell indexes are value-weighted, index assignment has asignificant effect on index weights and the extent of a stock’s ownership bypassive investors. The 950th largest stock at the end of May is more likely tobe included in the Russell 1000 and given a very small weight in the index,while the 1,050th largest stock is more likely to be included in the Russell

8 The number of stocks in the indexes can deviate from 1,000/2,000 for two reasons. First, the number of stocksin the indexes on the reconstitution date can deviate from 1,000/2,000 because of banding procedure. Second,stocks are added and/or removed from the indexes during the year because of acquisitions, bankruptcies, initialpublic offerings (IPOs), etc.

11



Figure 2Portfolio weights in the Russell 1000 and 2000 indexes by within-index ranking for 2013This figure plots the portfolio weights of the bottom-500 firms in the Russell 1000 index and the top-500 firmsin the Russell 2000 index for the end-of-June 2013. Observations are ordered by their within-index ranking suchthat rankings of 1 and 1,000 represent the firms with the largest and the 1,000th largest portfolio weight in theindex, respectively. The portfolio weights are given as a percentage.

2000 and given a much larger weight. For example, during our sample period,the average weight of the bottom-250 stocks in the Russell 1000 was 0.014%,whereas the average weight of the top-250 stocks in the Russell 2000 was anorder of magnitude larger at 0.145%. The difference in weights persists overa wide range around the cutoff, as observed in Figure 2. Because passive fundholdings mimic the weights in the underlying index to minimize tracking error,these differences in weights around the 1000/2000 cutoff significantly affectthe extent of a stock’s ownership by passive investors. For each dollar investedin a passive fund benchmarked to an index, a larger proportion is invested instocks at the top of the index than in stocks at the bottom.

The importance of index assignment for ownership by passive mutual fundsis illustrated in Figure 3, in which we rank stocks using their end-of-MayCRSP market capitalization and plot the average share of firms in the Russell2000 and average end-of-September ownership by passively managed funds.The sample in this figure contains the top-500 stocks of the Russell 2000 andbottom-500 stocks of the Russell 1000 for each year between 2007 and 2013, asdetermined using the end-of-June Russell-assigned weights within each index.By construction, the top panel of Figure 3 shows a smooth relation betweensize and ranking, but as shown in the middle panel, the largest stocks are in theRussell 1000; the smallest stocks are in the Russell 2000. In the intermediaterange around the cutoff, a stock’s probability of membership in the Russell 2000and its ranking are positively correlated, reflecting Russell’s banding policy thatlimits membership changes. The bottom panel of Figure 3 demonstrates that theownership of passive funds across rankings closely tracks the share of stocksassigned to the Russell 2000 with passive ownership being about 40% higherfor stocks in the Russell 2000.

The magnitude of the observed difference in passive ownership correspondsto the magnitude one would predict using estimates of the amount of passiveassets tracking each of the two indexes. For example, Chang, Hong, and

12



Figure 3Market capitalization, index assignment, and passive ownership by market capitalization rankings for thebottom-500 firms of Russell 1000 and top-500 firms of Russell 2000This figure plots the average end-of-May ln(market capitalization), the fraction of firm-year observations in theRussell 2000, and the passive mutual fund ownership (%) by ranking, where ranking is determined using end-of-May market capitalization, as reported in CRSP. The sample includes the bottom-500 firms of the Russell 1000and the top-500 firms of the Russell 2000, as determined using end-of-June Russell-assigned portfolio weightsfor each index. Average passive fund ownership is calculated as of September each year, and all averages arecalculated using bins of ten stocks (from 2007 to 2013). For the ownership panel, we scale the vertical axis toreport observations within 1 standard deviation of the sample mean of 9.4% (as reported in Table 1).

Liskovich (2015) estimate that $56.8 billion in assets were passively trackingthe Russell 2000 in 2010, which accounts for about 4.93% of the index’s totalmarket cap of $1,115 billion, while there was $137.1 billion of assets passivelytracking the Russell 1000, accounting for just 1.17% of the index’s total marketcap of $11,740 billion. Based on these estimates, assignment to the Russell2000, rather than to the Russell 1000, in that year would increase a stock’spassive institutional ownership by 3.76 percentage points, an increase thatis similar to the 3.4-percentage-point increase we detect in 2010 using ourmeasure of passive ownership. In practice, the realized differences in passiveownership we detect will be slightly smaller around the cutoff than predictedby this simple back-of-the-envelope calculation because passive investments

13



by some institutions, like pension funds, are not reported in the S12 mutualfund database.

The importance of index assignment for passive ownership is furtherhighlighted by examining the total ownership stake of the largest passiveinstitutions during our sample period: Vanguard, State Street, DFA, andBGI/Blackrock (the owners of iShares during our sample). For this, we usethe Thomson Reuters Institutional Holdings (13F) Database, which reports thetotal holdings, both passive and active, of each institution. On average, theownership stake of each of these four institutions is 30% higher for the 500firms at the top of the Russell 2000 relative to the bottom 500 firms of theRussell 1000, while the likelihood of each institution owning more than 5%of a firm’s shares is 60% higher and the likelihood of each institution being atop-five shareholder is 17% higher.

We find no evidence that index assignment is related to an increase inownership by actively managed funds and unclassified funds. We formallytest and demonstrate this in Section 2.3.

2.2 Identification strategy and empirical specificationFollowing AGK, we use an instrumental variable estimation strategy that relieson a stock’s Russell index assignment as a source of exogenous variation inpassive ownership. Because index assignment is determined by an arbitrary rulesurrounding the market capitalization of the 1,000th largest firm and firms’past index assignments, the higher passive ownership among stocks at thetop of the Russell 2000 relative to stocks at the bottom of the Russell 1000is plausibly exogenous after conditioning on the three factors that determinea firm’s index assignment: market capitalization, past index assignment, andwhether the firm’s market capitalization falls within a certain range of the1,000th largest firm. Therefore, we use inclusion in the Russell 2000 as aninstrument for ownership by passive funds in an estimation that controls for allfactors that determine stocks’ index assignments, including end-of-May marketcapitalization.

Unlike AGK, however, our sample period occurs after Russell’s switch tousing additional thresholds and past index assignments to determine a stock’syearly index assignment. We therefore augment their IV specification to includethree additional controls for each firm i and reconstitution year t (i.e., fromend-of-June year t to end-of-June year t +1): (1) an indicator for having an end-of-May market capitalization that ensures firm i will be “banded” by Russelland not switch indexes in reconstitution year t because the distance between itsmarket cap and the Russell 1000/2000 cutoff is less than 2.5% of the Russell3000E Index cumulative market cap, bandit , (2) an indicator for being in theRussell 2000 last reconstitution year t −1, R2000it−1, and (3) the interactionof these two indicators. These three additional controls capture the additional

14



criteria used by Russell beginning in 2007 when determining each firm’s indexassignment at the annual end-of-June reconstitution for year t .9

Specifically, we estimate the following activism event-level regression:

Yeit+1 =α+βPassive%it +N!

n=1

θn(ln(Mktcapit ))n +γ ln(F loatit )

+µ1bandit +µ2R2000it−1 +µ3(bandit ×R2000it−1)+δt +εeit

, (1)

where Yeit+1 is the outcome of interest for activism event e targeting firm i

in year t +1; Passive%it is the percentage of a firm’s shares held by passivelymanaged mutual funds at the end of the end of September in year t (i.e., in thefirst quarter after reconstitution in year t); Mktcapit is the end-of-May CRSPmarket capitalization of stock i in year t ; Floatit is the float-adjusted marketcapitalization calculated by Russell when setting the portfolio weights duringthe end-of-June reconstitution. To account for the possibility that Passive%might be correlated with the error term, εeit , because of omitted variable issuesdiscussed above, we instrument Passive% using R2000it , which is an indicatorequal to 1 if stock i is part of the Russell 2000 index in reconstitution yeart.We control for float-adjusted market capitalization because Russell uses itto compute portfolio weights and could be related to a firm’s stock liquidity,which might affect activism, and we include year fixed effects, δt , to ensurethat our estimates are identified using within-year variation. Finally, we clusterthe standard errors, εeit , at the firm level and scale Passive%it by its samplestandard deviation so that the point estimate of β reflects the change in Yeit+1 fora 1-standard-deviation increase in Passive%it . Our subsequent standard errorsare very similar if we instead cluster at the activist level.

Our IV estimation relies on the assumption that, after conditioning on thecriteria used to determine a stock’s index assignment, inclusion in the Russell2000 index is associated with an increase in Passive% (relevance condition) butdoes not directly affect our outcomes of interest except through its impact onownership by passive investors (exclusion restriction). We verify the relevancecondition below in our first-stage estimations, and the exclusion restrictionseems reasonable in that it is unclear why index inclusion would be directlyrelated to our outcomes of interest after robustly controlling for the factors thatdetermine index inclusion, including a firms’ end-of-May market capitalization.To bolster our assumption regarding the exclusion restriction, we will show in

9 These additional controls account for how banding affects the configuration of firms around the cutoff betweenthe Russell 1000 and 2000 indexes. In the post-banding period, stocks with better past stock returns will tendto remain in the Russell 2000 while stocks with worse past stock returns will tend to be kept in the Russell1000. The importance of including these additional controls in the post-banding period trade-offs is discussed inAppel, Gormley, and Keim (2015). Russell’s use of banding after 2006 also precludes the ability to use a fuzzyregression discontinuity estimation, as proposed for prebanding time periods by Wei and Young (2017), sincebanding eliminates any potential discontinuity in index assignment at the cutoff between the 1,000th and 1,001stlargest firms based on end-of-May market capitalization.

15



later tests that our instrument is not related to other factors that might plausiblyaffect activism outcomes, including analyst coverage, ownership by activelymanaged mutual funds, and targeted firm’s ex ante financial characteristics andperformance.

The use of R2000it as an instrument allows us to isolate an exogenoussource of variation in passive ownership. While nonindex funds that passivelyseek to deliver the performance of a benchmark portfolio have discretion overwhich stocks within the benchmark to hold, the instrumental variable neveruses such endogenous variation in passive ownership; the IV estimation onlyuses variation in ownership that is driven by a stock’s index assignment and thereshuffling of holdings by passively managed mutual funds seeking to minimizetheir tracking error. We do not use the actual portfolio weight or float-adjustedJune ranks of stocks as our instrument because this would introduce a potentiallyserious endogeneity concern.10

Our instrumental variable strategy differs from that used by Schmidt andFahlenbrach (2017), who use the endogenous switches from one index to theother as instruments for passive ownership. Making use of such switches is notpossible in our setting, because doing so would require firms to be targeted by anactivist both before and after switching indexes. The comparison of switchersversus nonswitchers also can be problematic, because stocks switching indexeslikely differ from stocks that remain in the same index.11

Our identification strategy instead relies on variation in passive ownershipby comparing stocks at the bottom of the Russell 1000 against stocks at the topof the Russell 2000, and because stocks in the two indexes will differ in theiraverage market capitalization, it will be important to show that our findings arenot sensitive to how we control for market capitalization. To control for firms’market capitalization, we restrict our sample to activism events that occur forthe bottom-500 stocks of the Russell 1000 and top-500 stocks of the Russell2000, and we include a robust set of controls for firms’ log market capitalization,ln(Mktcap), as measured using CRSP data, by varying the polynomial order N

we use to control for end-of-May market capitalization. In subsequent tests, weconfirm that our main findings are unchanged when using wider bandwidthsand qualitatively similar when using smaller bandwidths.

10 See Appel, Gormley, and Keim (2015, 2016) for more details. Chang, Hong, and Liskovich (2015), Mullins(2014), and Wei and Young (2017) also discuss this issue of why the actual weights or float-adjusted rankingsshould not be used as instruments or as part of a regression discontinuity estimation in the Russell 1000/2000setting.

11 The trade-offs of the different methodologies used in this identification setting are discussed in Appel, Gormley,and Keim (2015). Other recent papers use the Russell 1000/2000 cutoff as a source of variation in institutionalinvestors’ portfolio weights (Fich, Harford, and Tran 2015) and total institutional ownership, as measured in the13F filings, (see, e.g., Bird and Karolyi 2016; Boone and White 2015; Crane, Michenaud, and Weston 2016;Mullins 2014, among others). Appel, Gormley, and Keim (2016) and Schmidt and Fahlenbrach (2017) show thatthe observed increase in institutional ownership is driven by passive institutional investors, thus allowing one touse index assignment as an instrument for passive ownership.

16



2.3 First-stage estimationIn this section, we report estimates of our first-stage regression of passivemutual fund holdings on membership in the Russell 2000 index plus additionalcontrols. Specifically, we estimate

Passive%it =η+λR2000it +N!

n=1

χn(ln(Mktcapit ))n +σ ln(F loatit )

+φ1bandit +φ2R2000it−1 +φ3(bandit ×R2000it−1)+δt +ueit

,

(2)

where R2000it is a dummy variable equal to one if stock i is in the Russell2000 for reconstitution year t , and the other variables are as defined forEquation (1). In our initial tests, we also analyze other outcome measures,including the percentage of shares outstanding owned by all mutual funds; thepercentage of shares outstanding owned by actively managed funds; and the per-centage of shares outstanding owned by unclassified mutual funds. The model isestimated using all activism events from 2008 through 2014 that targeted firmswithin a bandwidth of 500 stocks around the Russell 1000/2000 threshold andincludes a second-order polynomial for ln(Mktcap).

The results, reported in Table 2, confirm that a targeted firm’s passiveownership structure is related to index assignment. So that the point estimates inTable 2 to align with the observed differences in ownership shown in Figure 3,we do not scale the ownership variables by their sample standard deviationsin these initial estimates. The first column shows that aggregate mutual fundownership is 6.3 percentage points higher for activist targets at the top of theRussell 2000, but the estimate is not statistically significant. Breaking mutualfund ownership into its different investment styles, however, we see that indexassignment is associated with the composition of a target’s ownership. The levelof passive ownership for targeted firms included in the Russell 2000 is about4.3 percentage points greater than the level of passive ownership observed fortargeted firms in the Russell 1000. The estimated coefficient is significant at the1% level (Column 2). There is no evidence that index assignment is related toownership of either actively managed mutual funds (Column 3) or unclassifiedfunds (Column 4).

In Table 3 we report estimates of the first-stage regression we use for theremainder of the paper in which we scale Passive% by its sample standarddeviation to better quantify the economic magnitude of the observed differencein ownership. We show that the estimated relation is robust to using higher-and lower-order polynomials to control for market cap, and find higher passiveownership at the top of the Russell 2000 relative to the bottom of the Russell1000 of about 1.1–1.2 standard deviations (Table 3, Columns 1–3). In all cases,the difference is statistically significant at the 1% level. Because our IV modelis just-identified, the IV estimation is median-unbiased and weak instruments

17



Table 2Impact of index assignment on mutual fund ownership

Percentage of firm’s common shares held by:

All mutual Passive Active UnclassifiedDependent variable = funds funds funds funds

(1) (2) (3) (4)

R2000 6.371 4.332∗∗∗ 1.851 0.188(4.047) (1.109) (3.779) (0.790)

Polynomial order, N 2 2 2 2Banding controls Yes Yes Yes YesFloat control Yes Yes Yes YesYear fixed effects Yes Yes Yes YesObservations 466 466 466 466R-squared .214 .447 .138 .125

This table reports estimates of a regression of mutual fund holdings on an indicator for membership in the Russell2000 index plus additional controls. Specifically, we estimate

Ownership%it =η+λR2000it +N!

n=1

χn[ln(Mktcapit )]n

+Xit +banding_controlsit +δt +ueit ,

where R2000it is a dummy variable equal to 1 if stock i is in the Russell 2000 index at end of June in yeart , Mktcapit is the CRSP market value of equity of stock i measured at May 31 in year t,N is the polynomialorder we use to control for ln(Mktcapit ), and δt are year fixed effects. The estimation includes an additionalcontrol for the natural log of the float-adjusted market value of equity on June 30 in year t , ln(Floatit ). Theestimate also includes additional banding controls: an indicator for having an end-of-May market capitalizationsufficiently close to the cutoff such that the firm will not switch indexes, bandit , an indicator for being in theRussell 2000 last year, R2000it−1, and the interaction of these two indicators. Ownership%it measures mutualfund ownership (in percent) for stock i at the end of September in year t . In this table we use four differentdefinitions for Ownership% for stock i: (1) the percentage of shares outstanding owned by all mutual funds(from S12 filings); (2) the percentage of shares outstanding owned by passive funds; (3) the percentage of sharesoutstanding owned by active mutual funds; and (4) the percentage of shares outstanding owned by unclassifiedmutual funds. The mutual fund classifications are defined in Section 1.1. The sample consists of all activismevents that target the top-500 firms in the Russell 2000 index and the bottom-500 firms of the Russell 1000 index(i.e., bandwidth = 500) over the 2008-2014 period for which we obtain holdings data from Thomson ReutersMutual Fund Holdings Database and which we match with data from the monthly CRSP file. The model isestimated using a polynomial order control for ln(Mktcap) of N =2. Standard errors, u, are clustered at the firmlevel and reported in parentheses. ***p<.01.

are unlikely to be a concern in our setting, especially given the strong first-stage estimates (Angrist and Pischke 2009). Additionally, the Kleibergen-PaapF stat on the excluded instrument exceeds 10, providing further confidence thata weak instrument is unlikely to be a concern (Stock, Wright, and Yogo 2002;Angrist and Pischke 2009).

3. Results: How Passive Investors Affect Activism by Other Investors

3.1 Likelihood of activismWe first examine whether passive ownership affects the likelihood of afirm being targeted by an activist. Theoretically, even if the presence ofpassive investors facilitates activism by lowering its cost or by increasingthe expected payoff of intervention, the effect of passive ownership on thefrequency of activism is ambiguous. By facilitating activism, the presence of

18



Table 3First-stage estimation for ownership by passively managed funds

Passive % scaled by itsDependent variable= sample standard deviation

(1) (2) (3)

R2000 1.103∗∗∗ 1.205∗∗∗ 1.216∗∗∗(0.302) (0.308) (0.316)

Polynomial order, N 1 2 3Banding controls Yes Yes YesFloat control Yes Yes YesYear fixed effects Yes Yes YesObservations 466 466 466R-squared .445 .447 .447

This table reports estimates of our first-stage regression of passive ownership onto an indicator for membershipin the Russell 2000 index plus additional controls. Specifically, we estimate

Passive%it =η+λR2000it +N!

n=1

χn[ln(Mktcapit )]n


where R2000it is a dummy variable equal to 1 if stock i is in the Russell 2000 index at end of June in year t ,Mktcapit is the CRSP market value of equity of stock i measured at May 31 in year t,N is the polynomial orderwe use to control for ln(Mktcapit ), and δt are year fixed effects. The estimation includes an additional controlfor the natural log of the float-adjusted market value of equity on June 30 in year t , ln(Floatit ). The estimate alsoincludes additional banding controls: an indicator for having an end-of-May market capitalization sufficientlyclose to the cutoff such that the firm will not switch indexes, bandit , an indicator for being in the Russell 2000 lastyear, R2000it−1, and the interaction of these two indicators. Passive%it is the percentage of shares outstandingowned by passively manged mutual funds, as defined in Section 1.1, for stock iat the end of September in year tscaled by its sample standard deviation. The sample consists of all activism events that target the top-500 firmsin the Russell 2000 index and the bottom-500 firms of the Russell 1000 index (i.e., bandwidth = 500) over the2008-2014 period for which we obtain holdings data from Thomson Reuters Mutual Fund Holdings Databaseand which we match with data from the monthly CRSP file. The model is estimated using a polynomial ordercontrols for ln(Mktcap) of N =1, 2, and 3. Standard errors, u, are clustered at the firm level and reported inparentheses. ***p<.01.

passive investors might increase its frequency. On the other hand, if managersinternalize this possibility and act to preempt activist campaigns (e.g., through areform of governance practices), because such campaigns are personally costlyfor the manager (Fos and Tsoutsoura 2014), the presence of passive investorsmight lower the frequency of activism. We might also observe a decline inthe likelihood of activism if the low-cost approaches to corporate governanceof passive investors improve firm-level governance and performance for somefirms (as found in AGK), thus (at least partially) negating the need for morecostly forms of activism by others.

Using our IV estimation, we find that the estimated effect of passiveownership on the likelihood of activism is statistically indistinguishable fromzero. The findings are reported in Table 4, where the dependent variable isan indicator for activism constructed using the SharkWatch database and thesample consists of all observations in the 500 bandwidth around the Russell1000/2000 cutoff during our sample period. The lack of an effect does notdepend on how we measure the occurrence of an activism event; omitting“13D only” activism events does not qualitatively change the findings, nor

19



Table 4Ownership by passive investors and the likelihood of a campaign

Dependent variable= Indicator for an activism campaign

(1) (2) (3)

Passive % −0.007 −0.015 −0.022(0.029) (0.029) (0.039)

Polynomial order, N 1 2 3Banding controls Yes Yes YesFloat control Yes Yes YesYear fixed effects Yes Yes YesObservations 6,803 6,803 6,803

This table reports estimates from a firm-level instrumental variable estimation used to identify the effect ofinstitutional ownership by passive investors on the likelihood of an activism event. Specifically, we estimate

Yeit+1 =α+β Passive%it =+N!

n=1

θn[ln(Mktcapit )]n

+Xit +banding_controlsit +δt +εeit ,

where Yeit+1 is an indicator for the likelihood of an activism event targeting firm i in year t +1, Passive%it is thepercentage of shares outstanding owned by passively managed mutual funds (as defined in Section 1.1) for stocki at the end of September in year t scaled by its sample standard deviation, Mktcapit is the CRSP market value ofequity of stock imeasured at May 31 in year t , and δt are year fixed effects. The estimation includes an additionalcontrol for the natural log of the float-adjusted market value of equity on June 30 in year t , ln(Floatit ). Theestimate also includes additional banding controls: an indicator for having an end-of-May market capitalizationsufficiently close to the cutoff such that the firm will not switch indexes, bandit , an indicator for being in theRussell 2000 last year, R2000it−1, and the interaction of these two indicators. We instrument Passive% in theabove estimation using R2000it , an indicator equal to 1 if firm i is part of the Russell 2000 index in year t . Thesample consists of the top-500 firms of the Russell 2000 index and the bottom-500 firms of the Russell 1000over the 2008–2014 period for which we obtain holdings data from Thomson Reuters Mutual Fund HoldingsDatabase and which we match with data from the monthly CRSP file. The model is estimated using polynomialorder controls N =1, 2, and 3 for ln(Mktcap). Standard errors, ε, are clustered at the firm level and reported inparentheses.

does using activism events, as defined by Brav et al. (2008) and Brav, Jiang,and Kim (2010). We analyze these possibilities further in Section 5.2.

The estimates in Table 4 are similar to those found in AGK, whichdocuments a small, but statistically significant, negative association betweenpassive ownership and the likelihood of activism. Using a sample from theyears 1998–2006, AGK reports a 1.6- to 2.0-percentage-point decline in thelikelihood of activism that is significant at the 5% level in one specificationand at the 10% level in two specifications (see Table 9 of AGK), while usinga sample period from 2008 to 2014, we find a 0.7- to 2.2-percentage-pointdecline in the likelihood of activism, though it is statistically insignificant (seeTable 4). AGK interpret these negative coefficients as evidence that passiveinvestors reducing the need for activism among some firms through their low-cost approaches to improving corporate governance. While there is no evidencethat passive ownership affects the likelihood of activism during our later sampleperiod, the similarity in coefficients across the two time periods also provideslittle evidence of any increase in activists’ likelihood of targeting firms withhigher passive ownership in recent years.

20



3.2 Type of activist campaignsWe now turn attention to whether passive ownership affects the types ofcampaigns initiated by activists. The presence of passive investors might affectthe type of activist campaigns, even absent a change in the frequency, if thefactors that affect an activist’s decision to initiate a campaign differ from thefactors that guide their strategic choices once a target is identified. In particular,activists might primarily rely on past underperformance and a significant scopefor future value improvements when choosing targets but then subsequentlytailor the nature of their campaigns depending on the firm’s ownership structureand the types of investors the activist needs to persuade. For example, if passiveinvestors are more receptive to governance- or board-related issues or if theirpresence lowers the coordination costs of tactics associated with seeking boardrepresentation, then activists might be more likely to seek board seats as partof their strategy to influence the target firm’s policy choices when passiveownership is higher. And, if passive investors tend to view policy changes, likeincreased dividends or debt, as either shortsighted or beyond their scope ofexpertise, then activists might be less likely to make such policy changes theonly goal of their campaign when passive ownership is higher. To analyze thispossible shift in the composition of campaigns, we now (and for the remainder ofthe paper) restrict the sample to those firms in the 500 bandwidth that experiencean activist event as defined by SharkWatch from 2008 through 2014.12

Table 5 reports the effects of passive ownership on each of the four groupsof activist campaigns described in Section 1.3 and summarized in Table 1. Wefind that higher passive ownership leads to an increase in campaigns seekingboard representation. Specifically, among firms targeted by an activist duringour sample period, a 1-standard-deviation increase in passive ownership isassociated with a 30-percentage-point increase in the likelihood of seekingboard representation (p-value < .05; Table 5, Column 1). The increase is sizablegiven that about 27.9% of campaigns seek board representation in our sample.The increase is robust to including higher-order polynomial controls for firm’send-of-May market cap; we observe a similar increase when including a second-or third-order polynomial control for market cap (p-values < .05; Table 5,Columns 2 and 3).

Given the lack of a relation between the overall likelihood of activism andpassive ownership reported in Table 4, the increased frequency of board-relatedcampaigns must be offset by a drop in the frequency of other types of campaigns.In Columns 4–6 of Table 5, we report results for these other types of campaigns;for brevity, we only report estimates that include a second-order polynomial

12 This sample restriction ensures that our coefficients reflect the effect of passive ownership on activists’ strategicchoices and campaign outcomes conditional on an actual activism event occuring. Limiting our sample to firmstargeted by activists does not introduce a selection issue since, as shown in Table 4, index assignment is notassociated with the likelihood of being targeted; see Angrist and Pischke (2009, pp. 64–66) for more detailsabout such selection concerns. In later robustness tests, we confirm that our findings are not sensitive to thissample restriction.

21



Table 5Ownership by passive investors and type of activist campaign

Maximize valuevia policy

Dependent variable = Seek board representation change Other 13D only

(1) (2) (3) (4) (5) (6)

Passive % 0.296∗∗ 0.355∗∗ 0.359∗∗ −0.084 −0.288∗ 0.017(0.135) (0.140) (0.140) (0.107) (0.169) (0.109)

Polynomial order, N 1 2 3 2 2 2Banding controls Yes Yes Yes Yes Yes YesFloat control Yes Yes Yes Yes Yes YesYear fixed effects Yes Yes Yes Yes Yes YesObservations 466 466 466 466 466 466

This table reports estimates of our instrumental variable estimation used to identify the effect of institutionalownership by passive investors on the type of activism campaign. Specifically, we estimate


n=1

θn[ln(Mktcapit )]n


where Yeit+1 is an indicator for the type of campaign for activism event e targeting firm i in year t +1, Passive%itis the percentage of shares outstanding owned by passively managed mutual funds (as defined in Section 1.1)for stock i at the end of September in year t scaled by its sample standard deviation, Mktcapit is the CRSPmarket value of equity of stock i measured at May 31 in year t , and δt are year fixed effects. The estimationincludes an additional control for the natural log of the float-adjusted market value of equity on June 30 in yeart , ln(Floatit ). The estimate also includes additional banding controls: an indicator for having an end-of-Maymarket capitalization sufficiently close to the cutoff such that the firm will not switch indexes, bandit , an indicatorfor being in the Russell 2000 last year, R2000it−1, and the interaction of these two indicators. The campaigntypes investigated in this table come from SharkWatch (Factset) and include an indicator for campaign that seeksboard representation (Columns 1-3), an indicator for campaigns that seek to maximize shareholder value byadvocating for specific corporate policy changes (Column 4), an indicator for all other campaign goals (Column5), and an indicator for the campaign only have a 13D filing with no stated intent (Column 6). We instrumentPassive% in the above estimation using R2000it , an indicator equal to 1 if firm i is part of the Russell 2000 indexin year t . The sample consists of all activism events that target the top-500 firms in the Russell 2000 index andthe bottom-500 firms of the Russell 1000 index (i.e., bandwidth = 500) over the 2008-2014 period for which wecan obtain holdings data from Thomson Reuters Mutual Fund Holdings Database and which we can match withdata from the monthly CRSP file. Standard errors, ε, are clustered at the firm level and reported in parentheses.*p<.10; **p<.05.

control for market cap. The increased likelihood of campaigns seeking boardrepresentation appears to largely come at the expense of campaigns classifiedas “other.” Specifically, a 1-standard-deviation increase in passive ownershipis associated with a 28.8-percentage-point drop (p−value < .10) in “other”campaigns, which is similar in magnitude to the observed increase in frequencyof board-related campaigns. We also find a negative point estimate for thelikelihood of a campaign seeking policy changes (Column 4), but the estimateis not statistically significant. The association between passive ownership andthe likelihood of a “13D only” filing is positive, but economically small andnot statistically significant (Column 6).

Overall, the results indicate that passive ownership leads to a meaningfulshift in the types of campaigns pursued by activists. Specifically, higherpassive ownership is associated with activists being more inclined to initiatecampaigns seeking to alter the balance of corporate control away from

22



incumbent directors. While activists are not necessarily seeking full controlof the board in such campaigns, an attempt to get “a chair at the metaphoricaltable where corporate strategy is set” (Kahan and Rock 2015, p. 192) representsan ambitious intervention on the part of the activist that holds the potential toaffect firms along virtually any dimension. The increase in campaigns seekingboard representation appears to be offset by a decrease in campaigns seekingincremental changes to firm policies through the use of shareholder nonbindingresolutions and exempt solicitations, among other means.

3.3 Likelihood of proxy fights and of obtaining board representationWhy is passive ownership associated with activist investors pursuing boardrepresentation? One possible explanation, as discussed above, is that passiveinvestors tend to focus on governance- and board-related issues. Knowing this,activists might tailor their campaigns to attract the support of the large passiveinstitutions. Another, but not mutually exclusive, possibility is that the presenceof passive investors might lower the cost of a common tactic used by activists towin board seats: proxy fights. Passive investors’ concentrated ownership stakesmight facilitate proxy battles by activists by reducing their coordination costsand ultimately increasing the chances of a favorable outcome. We analyze thispossibility by examining whether passive investors influence the likelihood ofactivists engaging in a proxy fight with management and whether activists aremore likely to obtain a successful outcome.

Proxy fights differ from many other activist tactics due to their considerablecost. These costs can be both direct (e.g., proxy solicitation services, legalfees) and indirect (e.g., effort) in nature. For example, an obviously importantaspect of a proxy fight is convincing other shareholders to vote for the dissidentdirectors. However, communication with other shareholders is complicated bythe fact that many hold shares in “street name” and cannot be easily identified.Thus, activists must hire proxy solicitation services. The costs associatedwith this are often considerable. For example, one study estimates the proxysolicitor fees alone cost activists $150,000 on average when the activist issues apreliminary or definitive proxy statement (Activist Insight 2014). Furthermore,activists often meet with other shareholders to convince them to vote for thedissident slate; Bebchuk (2007) notes that Red Zone LLC spent $950,000 fortravel alone in its proxy fight against Six Flags. Overall, Gantchev (2013)estimates the total average cost of a campaign ending in a proxy fight to beover $10 million.

Other types of activism, like supporting a particular shareholder proposal orseeking an exempt solicitation, are usually less costly. For example, the primarydirect cost for exempt solicitations is to “EDGARize” (i.e., format in accordancewith SEC guidelines) the filing, which costs about $100 (McRitchie 2012).An exempt solicitation features the dissident communicating, sometimes vialetter, with no more ten other shareholders about an upcoming director election,so indirect costs are likely minimal as well. Similarly, submitting shareholder

23



Table 6Ownership by passive investors and proxy fight likelihood and board seats sought

Dependent variable = Proxy fight Number of seats sought

(1) (2) (3) (4) (5) (6)

Passive % 0.312∗∗ 0.321∗∗∗ 0.310∗∗∗ 0.951∗∗ 0.941∗∗ 0.945∗∗(0.123) (0.122) (0.119) (0.455) (0.445) (0.444)

Polynomial order, N 1 2 3 1 2 3Banding controls Yes Yes Yes Yes Yes YesFloat control Yes Yes Yes Yes Yes YesYear fixed effects Yes Yes Yes Yes Yes YesObservations 466 466 466 466 466 466

This table reports estimates of our instrumental variable estimation used to identify the effect of institutionalownership by passive investors on the likelihood of a proxy fight and the number of board seats sought by anactivist. Specifically, we estimate


n=1

θn[ln(Mktcapit )]n


where Yeit+1 is the outcome of interest for activism event e targeting firm i in year t +1, Passive%it is thepercentage of shares outstanding owned by passively managed mutual funds (as defined in Section 1.1) for stocki at the end of September in year t scaled by its sample standard deviation, Mktcapit is the CRSP market value ofequity of stock i measured at May 31 in year t , and δt are year fixed effects. The estimation includes an additionalcontrol for the natural log of the float-adjusted market value of equity on June 30 in year t , ln(Floatit ). The estimatealso includes additional banding controls: an indicator for having an end-of-May market capitalization sufficientlyclose to the cutoff such that the firm will not switch indexes, bandit , an indicator for being in the Russell 2000last year, R2000it−1, and the interaction of these two indicators. The campaign outcomes investigated in thistable come from SharkWatch (Factset) and include an indicator for a proxy fight occuring (Columns 1-3) and thenumber of board seats sought by the activist (Columns 4-6). We instrument Passive% in the above estimationusing R2000it , an indicator equal to 1 if firm i is part of the Russell 2000 index in year t . The sample consistsof all activism events that target the top-500 firms in the Russell 2000 index and the bottom-500 firms of theRussell 1000 index (i.e., bandwidth = 500) over the 2008-2014 period for which we can obtain holdings datafrom Thomson Reuters Mutual Fund Holdings Database and which we can match with data from the monthlyCRSP file. Standard errors, ε, are clustered at the firm level and reported in parentheses. **p<.05; ***p<.01.

proposals for inclusion on a company’s proxy statement—provided that certainownership and procedural requirements are satisfied—comes at no cost (Briggs2007).

Consistent with the idea that passive investors lower the costs and/or increasethe expected benefits of launching a proxy fight, we find that higher passiveownership is associated with an increase in campaigns involving a proxy fight.Table 6 reports the results. Specifically, among firms targeted by an activist, a1-standard-deviation increase in passive ownership is associated with approx-imately a 30-percentage-point increase in the likelihood of a proxy fight, andthe point estimates are statistically significant at the 5% level (Columns 1–3).Relative to the average likelihood of a proxy fight, 18.9%, this corresponds toa sizable increase. Activists also appear to be more ambitious in the number ofboard seats they seek when passive ownership is higher. A 1-standard-deviationincrease in passive ownership is associated with about one additional seat beingsought (Columns 4–6; p-value < .05), relative to a sample average of 0.76 seatssought.

24



Next, we analyze the impact of passive ownership on proxy fight outcomes.There are four outcomes of proxy fights: settlement, withdrawal, vote in favor ofmanagement, and vote in favor of the activist. Settlements, which are generallya positive outcome for the activist, are typically associated with some boardrepresentation for the activist and occur when managers view an activist’scampaign as credible (Bebchuk et al. 2017). Withdrawals, on the other hand,occur when an activist anticipates defeat and withdraws their proxy fight. In oursample, approximately 36.5% of proxy fights end in a settlement and 28.6%end with a withdrawal by the activist. Of the remaining 34.9% of proxy fightsin our sample that reach a shareholder vote, roughly half of the votes are wonby management.

We find that passive ownership is associated with an increase inactivists obtaining concessions and board representation via settlements, andno evidence that these additional campaigns are less likely to succeed.A 1-standard-deviation increase in passive ownership is associated withapproximately a 16- to 20-percentage-point increase in the likelihood that anactivist campaign results in a proxy settlement, and the estimates are statisticallysignificant at the 5% level (Table 7, Columns 1–3). These findings suggest thatthe ex ante probability that an activist will win a proxy fight is higher whenpassive investors hold a larger share of the firm’s stock. We find no evidencethat passive ownership is associated with differences in the probability that anactivist anticipates defeat and withdraws their proxy fight (Column 4) or in therates at which activists or managers win proxy fights that go to a shareholdervote (Columns 5 and 6).

We also find evidence that passive ownership is associated with an increasein favorable settlement outcomes. A 1-standard-deviation increase in passiveownership is associated with about 0.25–0.35 additional seats won in campaignsthat end with a settlement (Table 7, Columns 7–9). The increase, which isstatistically significant at the 10% level in two of the three specifications,corresponds to an increase of about 200%–300% relative to the sample averageof 0.11 board seats being won.

Combined, the increase in campaigns with proxy fights and favorable proxyfight outcomes for activists are consistent with the presence of passive investorslowering activists’ cost of initiating a proxy fight and increasing the expectedbenefits of launching a proxy battle. Furthermore, the shift toward more costlyboard-related campaigns involving proxy fights, accompanied by no changein the overall likelihood of activism (see Section 3.1), is consistent with thepossibility that passive ownership substitutes for activism in situations whereless costly forms of monitoring are effective (as found in AGK) but facilitatesactivism in situations where more costly forms of engagement are required.

If activists nominate different types of directors or if passive investorshave different director preferences, we might also observe a change inthe composition of boards after campaigns targeting firms with morepassive ownership. In untabulated findings using BoardEx data on director

25



Table 7Ownership by passive investors and proxy fight outcomes

Activist Firm Number of board seatsDependent variable = Proxy fight settlement wins wins Withdrawn won in settlement

(1) (2) (3) (4) (5) (6) (7) (8) (9)

Passive % 0.164∗∗ 0.200∗∗ 0.192∗∗ 0.036 0.044 0.042 0.266 0.351∗ 0.345∗(0.083) (0.088) (0.086) (0.047) (0.047) (0.047) (0.182) (0.196) (0.197)

Polynomial order, N 1 2 3 2 2 2 1 2 3Banding controls Yes Yes Yes Yes Yes Yes Yes Yes YesFloat control Yes Yes Yes Yes Yes Yes Yes Yes YesYear fixed effects Yes Yes Yes Yes Yes Yes Yes Yes YesObservations 466 466 466 466 466 466 466 466 466

This table reports estimates of our instrumental variable estimation used to identify the effect of institutionalownership by passive investors on proxy fight outcomes. Specifically, we estimate


n=1

θn[ln(Mktcapit )]n


where Yeit+1 is an indicator for the proxy fight outcome for activism event e targeting firm i in year t +1,Passive%it is the percentage of shares outstanding owned by passively managed mutual funds (as defined inSection 1.1) for stock i at the end of September in year t scaled by its sample standard deviation, Mktcapit isthe CRSP market value of equity of stock i measured at May 31 in year t , and δt are year fixed effects. Theestimation includes an additional control for the natural log of the float-adjusted market value of equity on June30 in year t , ln(Floatit ). The estimate also includes additional banding controls: an indicator for having an end-of-May market capitalization sufficiently close to the cutoff such that the firm will not switch indexes, bandit ,an indicator for being in the Russell 2000 last year, R2000it−1, and the interaction of these two indicators. Theproxy fight outcomes investigated in this table come from SharkWatch (Factset) and include an indicator for aproxy settlement between the firm and the activist (Columns 1-3), the activist winning the vote in a proxy fight(Column 4), the firm winning the vote in a proxy fight (Column 5), the activist withdrawing the proxy fight beforea vote occurs (Column 6), and the number of board seats won in a settlement (Columns 7-9). We instrumentPassive% in the above estimation using R2000it , an indicator equal to 1 if firm i is part of the Russell 2000 indexin year t . The sample consists of all activism events that target the top-500 firms in the Russell 2000 index andthe bottom-500 firms of the Russell 1000 index (i.e., bandwidth = 500) over the 2008-2014 period for which wecan obtain holdings data from Thomson Reuters Mutual Fund Holdings Database and which we can match withdata from the monthly CRSP file. Standard errors, ε, are clustered at the firm level and reported in parentheses.*p<.10; **p<.05.

characteristics, we find that higher passive ownership is associated with newlyappointed directors who are more likely to be independent (which aligns wellwith passive investors’ stated preference for independent directors and theearlier findings of AGK) and well connected (as measured by the numberof other boards of listed firms on which directors have previously served).Passive ownership is also associated with differences in the characteristicsof departing directors. When passive ownership is higher, departing directors(after an activist campaign) tend to be older and tend to have spent more timeat the company. Overall, these findings suggest that boards likely compriseof younger, more independent, shorter-tenured, and better-connected directorsfollowing activist campaigns that target firms with greater passive ownership.

3.4 Activists’ likelihood of success in obtaining nonboard objectivesIn this section, we test whether passive ownership is associated withactivists’ likelihood of success in dimensions besides proxy fights and board

26



Table 8Ownership by passive investors and nonproxy fight activist outcomes

Corporate control and governance outcomes Other corporate policy outcomes

Acquired Acquired Removed Capital[by third [by Merger takeover Increased structure

Dependent variable = party] activist] blocked defense payouts change Spinoff(1) (2) (3) (4) (5) (6) (7)

Passive % 0.110∗ 0.060∗ −0.174∗ 0.058 0.051 −0.022 −0.026(0.066) (0.032) (0.094) (0.035) (0.038) (0.022) (0.047)

Polynomial order, N 2 2 2 2 2 2 2Banding controls Yes Yes Yes Yes Yes Yes YesFloat control Yes Yes Yes Yes Yes Yes YesYear fixed effects Yes Yes Yes Yes Yes Yes YesObservations 466 466 466 466 466 466 466

This table reports estimates of our instrumental variable estimation used to identify the effect of institutionalownership by passive investors on non-proxy fight activist outcomes. Specifically, we estimate


n=1

θn[ln(Mktcapit )]n


where Yeit+1 is an indicator for the outcome of activism event e targeting firm i in year t +1, Passive%it isthe percentage of shares outstanding owned by passively managed mutual funds (as defined in Section 1.1)for stock i at the end of September in year t scaled by its sample standard deviation, Mktcapit is the CRSPmarket value of equity of stock i measured at May 31 in year t , and δt are year fixed effects. The estimationincludes an additional control for the natural log of the float-adjusted market value of equity on June 30 in yeart , ln(Floatit ). The estimate also includes additional banding controls: an indicator for having an end-of-Maymarket capitalization sufficiently close to the cutoff such that the firm will not switch indexes, bandit , an indicatorfor being in the Russell 2000 last year, R2000it−1, and the interaction of these two indicators. The campaignoutcomes investigated in this table come from SharkWatch (Factset) and include an indicator for whether thefirm is acquired by a third party (Column 1), is acquired by the activist (Column 2), has a merger blocked(Column 3), removes a takeover defense (Column 4), increases its payout policy (Column 5), makes a changeto its capital structure (Column 6), or does a spinoff or divestiture (Column 7). We instrument Passive% in theabove estimation using R2000it , an indicator equal to 1 if firm i is part of the Russell 2000 index in year t . Thesample consists of all activism events that target the top-500 firms in the Russell 2000 index and the bottom-500firms of the Russell 1000 index (i.e., bandwidth = 500) over the 2008-2014 period for which we can obtainholdings data from Thomson Reuters Mutual Fund Holdings Database and which we can match with data fromthe monthly CRSP file. Standard errors, ε, are clustered at the firm level and reported in parentheses. *p<.10.

representation. To do this, we create indicator variables to flag campaigns inwhich the activist was successful in obtaining specific policy outcomes relatedto corporate control and governance (e.g., acquisitions and takeover defenses)and corporate policies (e.g., increased payouts, capital structure changes, andspinoffs). For brevity, we continue to restrict our analysis to estimations thatinclude a second-order polynomial control for ln(Mktcap).

Our results, reported in Table 8, indicate that passive investors have asignificant effect on the likelihood of activists achieving outcomes relatedto changes in corporate control. For example, greater passive ownership isassociated with an increased likelihood that activists successfully push foran acquisition of the target. Among firms targeted by activists, a 1-standard-deviation increase in passive ownership is associated with an 11-percentage-point increase in the likelihood the activist successfully seeks and obtains anacquisition by a third party (Column 1) and a 6-percentage-point increase in

27



the likelihood the activist seeks and acquires the target firm itself (Column 2).Both estimates are statistically significant at the 10% level and consistent withpassive investors being inclined to support acquisitions where shareholders arebeing offered a premium. Given the potentially large value implications of theseoutcomes, increased activist success also might be the result of an increasedthreat of a proxy fight should the activist face resistance from managers. Greaterpassive ownership, however, is also associated with a decline in the likelihoodan activist is able to successfully block a merger or agitate for a higher price ina proposed merger (Column 3). One potential explanation for this latter findingis that passive investors often hold significant ownership stakes in both theacquirer and the target, thus mitigating their incentive to support a higher price,irrespective of whether such a price increase might be beneficial to the targetor other involved parties.

We also find suggestive evidence that greater passive ownership is associatedwith increased success by activists in removing takeover defenses. Reform ofgovernance practices, including the removal of takeover defenses, is a commongoal of activists. We find that a 1-standard-deviation increase in passiveownership is associated with a 6-percentage-point increase in the likelihood ofthe firm removing a takeover defense (Table 8, Column 4), though the estimateis not statistically significant at conventional levels (p−value = .101). Thisfinding is consistent with recent evidence that passive investors tend to opposetakeover defenses (AGK 2016).

An increased presence of passive investors, however, does not appear tohave an effect on activists’ ability to instigate changes to corporate policiesunrelated to governance or corporate control. Specifically, we find less evidencethat passive ownership is related to success in obtaining an increase in payouts(Table 8, Column 5), changing the capital structure (Column 6), or facilitating aspinoff (Column 7). The nonfindings are consistent with passive investors beingless inclined to support such changes, because they view them as either betterleft to the discretion of managers and boards (AGK 2016) or short sighted, asargued by Larry Fink, CEO of Blackrock (Benoit 2014a).

3.5 Activists’ likelihood of using other nonproxy fight tacticsIn practice, activists can employ a combination of tactics besides nominatinga slate of directors and initiating a proxy fight. The most common alternativetactic is to write a letter to the board and other shareholders. Other tacticscan include initiating a lawsuit, obtaining a vote on a precatory shareholderproposal, pushing for a vote on a binding proposal, and seeking reimbursementfor expenses occurred.

Passive ownership has less effect on these other tactics pursued by activists.Table 9 reports these results. While passive ownership is positively associatedwith an increase in the likelihood an activist initiates a lawsuit, which isgenerally considered a more hostile tactic, the point estimate is not statisticallysignificant at conventional levels (Table 9, Column 1). We find that passive

28



Table 9Ownership by passive investors and other activist tactics

Letter to SeekSH Precatory Binding reimburse-

Dependent variable = Lawsuit (nonproxy) proposal proposal ment(1) (2) (3) (4) (5)

Passive % 0.105 −0.259∗ −0.111 0.038 0.188∗∗(0.070) (0.151) (0.113) (0.073) (0.091)

Polynomial order, N 2 2 2 2 2Banding controls Yes Yes Yes Yes YesFloat control Yes Yes Yes Yes YesYear fixed effects Yes Yes Yes Yes YesObservations 466 466 466 466 466

This table reports estimates of our instrumental variable estimation used to identify the effect of institutionalownership by passive investors on the tactics of activism campaigns. Specifically, we estimate


n=1

θn[ln(Mktcapit )]n


where Yeit+1 is an indicator for whether certain activism tactics were used in event e targeting firm i in yeart +1, Passive%it is the percentage of shares outstanding owned by passively managed mutual funds (as definedin Section 1.1) for stock i at the end of September in year t scaled by its sample standard deviation, Mktcapitis the CRSP market value of equity of stock i measured at May 31 in year t , and δt are year fixed effects. Theestimation includes an additional control for the natural log of the float-adjusted market value of equity on June30 in year t , ln(Floatit ). The estimate also includes additional banding controls: an indicator for having an end-of-May market capitalization sufficiently close to the cutoff such that the firm will not switch indexes, bandit ,an indicator for being in the Russell 2000 last year, R2000it−1, and the interaction of these two indicators. Theactivism tactics investigated in this table come from SharkWatch (Factset) and include an indicator for initiatinga lawsuit (Column 1), writing a nonproxy letter to shareholders or the board (Column 2), offering a precatoryshareholder proposal (Column 3), pushing for a vote on a binding proposal (Column 4), or seeking reimbursement(Column 5). We instrument Passive% in the above estimation using R2000it , an indicator equal to 1 if firm i ispart of the Russell 2000 index in year t . The sample consists of all activism events that target the top-500 firmsin the Russell 2000 index and the bottom-500 firms of the Russell 1000 index (i.e., bandwidth = 500) over the2008-2014 period for which we can obtain holdings data from Thomson Reuters Mutual Fund Holdings Databaseand which we can match with data from the monthly CRSP file. Standard errors, ε, are clustered at the firm leveland reported in parentheses. *p<.10; **p<.05.

ownership is associated with fewer letters to shareholders (p<0.10) but littleevidence of a change in the likelihood of precatory shareholder proposals, bothof which are relatively lower-cost tactics (Columns 2 and 3). We also find littleassociation between passive ownership and activists pushing a binding proposal(Column 4).

Passive ownership, however, is associated with an increase in activistsseeking reimbursement from the firm. Such requests are often made in expensiveproxy fights,13 and consistent with the earlier observed increase in proxy fights,we find that a 1-standard-deviation increase in passive ownership is associatedwith an 18.8-percentage-point increase in the likelihood the activist seeks

13 For example, an article by Sullivan & Cromwell LLP, a law firm that provides legal advice to corporate boards,notes the following: “If an investor has only threatened a contest but has not prepared or filed a proxy statementwith the SEC or begun soliciting proxies, it may not have incurred significant expenses and may not be concernedabout reimbursement. For an investor who has engaged in a protracted proxy fight with a company, the investor’sexpenses are likely to be substantial, and reimbursement will become a bigger issue.” See Aquila (2015, p. 27).

29



reimbursement. The magnitude of this increase in reimbursement requestsis sizable given that only 9.9% of campaigns in our sample make suchrequests. This finding provides additional evidence that activists undertake moreexpensive campaigns in the presence of higher passive ownership.

3.6 Market perceptions and long-term performanceFinally, we analyze the impact of passive ownership on the market’s responseto the announcement of an activist’s campaign, and the impact of that campaignon a target’s long-term accounting performance. Researchers have documenteda positive average market response to activist campaigns (for a summary,see Brav, Jiang, and Kim 2015a), which reflect both investors’ perceptionsregarding the value impact of the activist’s proposed changes and the perceivedlikelihood the activist will succeed in obtaining those changes. If passiveownership increases either the likelihood of an activist succeeding, as suggestedabove, or an activist’s ability to seek more value-enhancing changes in a targetedfirm, we might expect to find a larger positive market response when passiveownership is higher. And to the extent the changes sought by activists improvetargeted firms’ long-term accounting performance (see, e.g., Brav et al. 2008;Bebchuk, Brav, and Jiang 2015), we might also expect to find that greaterpassive ownership leads to a larger positive impact on long-term performance.

Consistent with passive investors improving activists’ ability to enhanceshareholder value, we find that higher passive ownership is associated witha larger positive market response at the time of an activist’s engagement. Tomeasure market response, we compute cumulative abnormal returns (CARs)using the four-factor Fama and French model in a 20-day window aroundthe public announcement of an activist campaign and use this as the outcomevariable in our instrumental variable estimation. We choose a [−10, 10] windowbecause Brav, Jiang, and Kim (2015a) shows that most of the price responsefrom activist events occurs in this period. The average market response inour sample is 4.5% (standard deviation = 16.8%), which is similar to theaverage market response found in other papers (see Brav, Jiang, and Kim2015a). As shown in Table 10, a 1-standard-deviation increase in passiveownership is associated with about an 11- to 12-percentage-point increase in theaverage CAR at time of announcement. The estimate is economically large andconsistent with the higher abnormal returns associated with hostile campaignsand campaigns involving the target’s sale, both of which occur more oftenwhen passive ownership is higher; Brav et al. (2008) document that abnormalreturns are, on average, 3.76 and 8.54 percentage points higher for campaignsassociated with hostile tactics and the attempted sale of the target, respectively.The findings are not sensitive to the way one adjusts for risk when calculatingCARs; we find similarly large estimates when we instead use the CAPM or athree-factor Fama and French models.

Our findings regarding the effect of passive ownership on market responsesto campaign announcements, however, are not as statistically robust as those

30



Table 10Ownership by passive investors and abnormal returns

Dependent variable = CAR(-10,10)

(1) (2) (3)

Passive % 0.112 0.115 0.123∗(0.072) (0.071) (0.073)

Polynomial order, N 1 2 3Banding controls Yes Yes YesFloat control Yes Yes YesYear fixed effects Yes Yes YesObservations 410 410 410

This table reports estimates of our instrumental variable estimation used to identify the effect of institutionalownership by passive investors on the cumulative abnormal return around the announcement of an activistcampaign. Specifically, we estimate


n=1

θn[ln(Mktcapit )]n


where Yeit+1 is the cumulative abnormal return (CAR) estimated using a 4-factor Fama and French model forthe 20-day window around the announcement of an activism event targeting firm i in year t +1, Passive%it is thepercentage of shares outstanding owned by passively managed mutual funds (as defined in Section 1.1) for stocki at the end of September in year t scaled by its sample standard deviation, Mktcapit is the CRSP market value ofequity of stock imeasured at May 31 in year t , and δt are year fixed effects. The estimation includes an additionalcontrol for the natural log of the float-adjusted market value of equity on June 30 in year t , ln(Floatit ). Theestimate also includes additional banding controls: an indicator for having an end-of-May market capitalizationsufficiently close to the cutoff such that the firm will not switch indexes, bandit , an indicator for being in theRussell 2000 last year, R2000it−1, and the interaction of these two indicators. We instrument Passive% in theabove estimation using R2000it , an indicator equal to 1 if firm i is part of the Russell 2000 index in year t . Thesample consists of the top-500 firms of the Russell 2000 index and the bottom-500 firms of the Russell 1000over the 2008–2014 period for which we obtain holdings data from Thomson Reuters Mutual Fund HoldingsDatabase and which we match with data from the monthly CRSP file. The model is estimated using polynomialorder controls N =1, 2, and 3 for ln(Mktcap). Standard errors, ε, are clustered at the firm level and reported inparentheses. *p<.10.

related to proxy fights and board representation. The estimated coefficients arestatistically significant at the 10% level in only one of the three specifications(e.g., the p-values in Columns 1 and 2 are .121 and .102, respectively), andwhile we find similarly positive point estimates using a longer post-eventwindow (e.g., [−20, 20]), those estimates are not statistically significant. Theweaker evidence for abnormal returns might seem surprising given passiveownership’s strong association with an increased used of proxy fights byactivists (see Table 6) and existing evidence of a large abnormal price reactionto the announcement of proxy fights (e.g., Fos 2017). One possible explanationis that the announcement of an activist campaign need not coincide with theannouncement of a proxy fight; the declaration of a proxy fight might not occuruntil after a campaign is announced and initial discussions with managementfail to resolve the activists’ concerns. A second explanation might be thatpassive ownership facilitates activists’ willingness to pursue proxy fights inmore marginal cases where the expected gains from such an aggressive tacticare lower.

31



Determining whether higher passive ownership is associated with greaterimprovements in long-term accounting performance following an activistcampaign is more challenging. The impact of activist campaigns may takemany years to manifest themselves fully into firm performance (Bebchuk, Brav,and Jiang 2015) and might be confounded and/or offset by other subsequentfirm-specific events. Further, attempts to analyze the passive investors’ impacton long-term performance are likely to suffer from selection bias since firmswith higher passive ownership are more likely to be subsequently acquired(Table 8), thus removing them from the sample. If such attrition is associatedwith a positive improvement in performance, as found in Brav, Jiang, and Kim(2015b), estimates of the impact on long-term performance would be negativelybiased (Bebchuk, Brav, and Jiang 2015). Consistent with this possibility, inunreported tests we find that more than a third of targeted firms exit thesample within two years of an activist campaign, and we find little evidenceof an association between passive ownership and the long-term accountingperformance following activist campaigns.

Our earlier findings, however, are suggestive that passive ownership is likelyto be positively associated with improvements in long-term performance.Bebchuk et al. (2017) find campaign-related increases in payouts toshareholders, long-term performance, and favorable delistings tend to beconcentrated around target firms that reach a settlement with the activist. Asshown earlier in Table 7, we find that such proxy fight settlements are muchmore likely when passive ownership is higher. Our earlier findings also showthat passive ownership is associated with an increased likelihood of target beingacquired, which are also events that have large, positive value implication fortarget shareholders.

4. Discussion of Possible Mechanisms

While our empirical setting provides exogenous variation in the concentrationof passive holdings, thus allowing us to identify the effect of passiveownership on activism outcomes, it does not provide exogenous variationin the potential mechanisms by which passive ownership might influenceactivists. For example, does passive ownership matter because it is increasesownership concentration, which, in turn, lowers coordination costs? Or doespassive ownership matter because of something particular to passive investors,including their long-term investment strategy and focus on governance andcontrol issues?

4.1 Shared desire for improved governance and tacit alliancesOur evidence suggests that passive investors matter because of their long-terminvestment strategies and focus on governance issues. If the effect of passiveownership were working solely through an increase in ownership concentrationand reduced coordination costs, we would expect to find increased successes for

32



activists in multiple dimensions, including their efforts to increase payouts andadjust capital structures. Instead, we only find that increased activists’ successesin areas that passive investors view as beneficial for their long-term interests; inparticular, effective boards, good governance, and a strong market for corporatecontrol. And activist campaigns associated with greater passive ownership aremore focused on board quality and representation and, thus, are more alignedwith passive investors’ proxy voting guidelines (see, e.g., the appendix of AGKfor the proxy voting guidelines of three large passive investors). As noted in arecent survey of long-term investors by the corporate governance firm MorrowSodali, poor governance practices are the biggest reason long-term investorssupport activist proposals.

Anecdotal evidence suggests the influence of passive investors on activistoutcomes is tacit, rather than through the creation of formal alliances.According to the chief legal officer of a hedge fund involved in multipleactivist campaigns, the fund begins talking with other shareholders wheninitial private conversations with the targeted company fail to yield resultsand the fund decides to go public with its campaign. When speaking to otherowners, the fund typically first reaches out to the proxy-voting departmentof the largest institutional shareholders in an attempt to educate them aboutits views. A hoped-for outcome of such communications is that some ofthese institutions will be persuaded by the activists’ arguments and signaltheir likely support in a proxy fight or privately communicate their supportto the targeted firm’s management and directors, as occurred in ValueAct’scampaign against Microsoft in 2012–2013 (Benoit and Grind 2015). The headsof the governance divisions at three large passive institutions confirm that suchconversations occur (including an example of Nelson Peltz flying to Londonfor such a meeting) and that their governance committee is the first point ofcontact for an activist because of their large ownership stakes and becauseactivists know the passive institutions will not be selling their shares beforeany proxy vote. One governance director also confirms that his institutionmight then privately communicate its views to managers and directors, andthat these communications can influence managers to reach a settlement withthe activists.14 Martin Lipton (2017), a lawyer who has represented companiesfacing activists, tells a similar story: as he notes, traditional institutionalinvestors do not actively join activist campaigns, but when supportive, theydo let the lead activist know that “it can count on their support in a proxy fight.”

14 One such example of a passive investor potentially influencing management to settle occurred in Third Point’scampaign against Sotheby’s in 2014, which is included in our sample. Court documents revealed that Sotheby’swas preparing for a tough vote after Blackrock, a major shareholder in Sotheby’s, communicated to Sotheby’s topmanagement that Third Point would win, and the day before the shareholder vote, Sotheby’s reached a settlementagreement with Third Point, appointing three new directors and removing a poison pill. See Biers (2014).

33



4.2 Characteristics of target firms, activist ownership stakeAnother mechanism by which greater passive ownership might affect activismoutcomes is by influencing the financial characteristics and average profitabilityof firms targeted by activists or the size of the ownership stakes taken byactivists when engaging. For example, the documented positive associationbetween passive ownership and proxy fights could reflect greater willingnessby activists, when aligned with passive institutions, to target less profitablefirms with intransigent managers and boards. The presence of passive investorsmight also influence the size of the ownership stake taken by activists which, inturn, could affect their subsequent strategic choices. For example, activistsmight take smaller positions in the targeted firm if they can expect thesupport of large passive institutions. Alternatively, activists might take largerpositions if this expected support increases the likelihood of success for theircampaign.

We find no evidence, however, that the amount of passive mutual fundownership is associated with either the ownership stakes taken by activists orcharacteristics of targeted firms. In particular, the size of the activists’ ownershipstakes (as measured by the percentage of holdings) is unrelated to passiveownership, and passive ownership does not have a statistically significantassociation with target firms’ cash holdings, dividend yield, leverage ratio,level of capital expenditures or research and development (R&D) expenses,return on assets, Tobin’s q, or stock return in the year prior to being targeted.Moreover, using the same matching approach as Brav et al. (2008), we findthat firms targeted by activists in our sample have lower q, growth, payouts,and cash, but higher leverage than the set of matched firms. This matches thefindings of Brav et al. (2008), and holds regardless of whether the target hashigher or lower passive ownership.

4.3 Exclusion restriction and other possible mechanismsAn underlying assumption of our identification strategy is that Russell indexinclusion affects activism outcomes only through its impact on the extent of astock held by passive investors. If index inclusion, however, is also related tothe extent of ownership by actively managed funds benchmarked to the indexor the extent of analyst coverage, then these could be alternative explanationsfor our findings if either analyst coverage or the extent of ownership by activelymanaged funds affect activists’ choices.

However, as shown earlier in Table 2, our main first-stage estimation yieldsno evidence that Russell index assignment is associated with the ownershiplevels of either actively managed or unclassified funds. Moreover, this findingis robust to varying the polynomial order of controls for Mktcap (see Table A.2).The point estimates for both active ownership and unclassified ownership areeconomically small and not statistically significant in any of the specifications.

Our findings also do not appear to be driven by a relation between Russellindex assignment and analyst coverage, which could represent another possible

34



violation of our underlying exclusion restriction. We find no evidence that beingassigned to the Russell 2000 is associated with a different level of analystcoverage after conditioning on index determinants. Table A.3 reports theseresults.

5. Robustness Checks

5.1 Robustness to alternative sampling choices, controls, and placebotests

In our main analysis, we select the sample to be the 500 stocks with thesmallest portfolio weights in the Russell 1000 and the 500 stocks with thelargest portfolio weights in the Russell 2000. Our findings, however, are notsensitive to this choice. This is shown in Figure A.1, in which we plot the pointestimates and 95th percentile confidence intervals when varying the bandwidthbetween 250 and 750 firms and using a second-order polynomial control forln(Mktcap); estimates are reported for the specifications in Tables 3–10. Theestimates are relatively similar across the entire range of bandwidths.

Our findings are also not sensitive to instead using end-of-May market capsto determine the sample of stocks each year. In particular, we can rank stocksbased on their end-of-May market cap, calculated with data from CRSP, andselect the sample for each year using firms ranked 500th through 1,500th inthat year. An advantage of this latter approach is that it eliminates the riskthat Russell’s float-adjusted reweighting of stocks within an index affects ourfindings. A disadvantage of this approach, however, is that we are no longernecessarily comparing the very bottom firms of the Russell 1000 against thevery top firms of the Russell 2000, which is where we would expect to findthe biggest difference in passive ownership (and, hence, outcomes) to occur.This sampling choice, however, minimally affects our IV estimates. While thefirst-stage estimates are expectedly smaller in magnitude and noisier when weuse end-of-May market caps to rank stocks and select our sample each year(coefficient = 0.83; t-stat = 2.75), the IV estimations are largely unchanged (seeTable A.4).

Including observations where no activism event occurs also does not affectour findings. To illustrate this, we repeat our analysis after coding activismoutcomes as zeros for firm-years with no activism event; for example, a firm thatis not targeted by an activist that year is also not being subjected to an activism-driven proxy fight that year. While the magnitude and interpretation of thecoefficients now reflect the effect of passive ownership on activism outcomesfor all firms, even those not subject to a campaign, the findings are similar(see Table A.5). We do not analyze announcement returns in this test since anannouncement return is undefined for firms not subject to an activism campaign.

Our findings are also largely unaffected if we add controls to account for astock’s liquidity. If an increase in passive ownership improves a stock’s liquidity,then this could be an additional mechanism by which passive ownership affects

35



activism outcomes. However, including controls for both Amihud’s measure ofilliquidity and bid-ask spread minimally affects our findings (see Table A.6).Our findings are also robust to instead controlling for the ratio of float-adjustedmarket capitalization to total market capitalization rather than controlling forln(Float) like in our main analysis. These results suggest that a difference orchange in liquidity is unlikely to be the key mechanism by which higher passiveownership affects the strategic choices of activists and the outcomes of theircampaigns.

Our findings are also robust to including controls for whether a company’sstock switched indexes. If index switchers differ in other dimensions andrepresent a disproportionate share of either index, this could affect our earlierestimates. However, all of the findings are robust to the inclusion of controlsfor whether a firm’s stock moved from the Russell 1000 to the Russell 2000that year, and vice versa.

Our findings are also robust to how we define an activist campaign. Excludingactivist campaigns where we only observe the filing of a 13D, but no subsequentinformation on the tactics employed or changes sought by the activist, whichoccurs in 67 of our 466 activist campaigns, does not affect our findings.Table A.7 shows this. Our findings are also robust to combining activistcampaigns that occur within the same year. In our main analysis, we treateach activist event reported in SharkWatch as a separate campaign. However,collapsing multiple activist events that occur in the same year into one combinedcampaign does not qualitatively affect our findings.

Clustering our standard errors at the activist level, rather than at the campaignlevel, does not qualitatively affect our findings. Table A.8 repeats our mainfindings, but, instead, we cluster the standard errors at the activist level.

Finally, in further support that our findings are not driven by specificationerror, we do not find an association between passive ownership and ouroutcomes of interest in placebo IV or reduced-form tests that use alternativethresholds. For example, if we restrict the sample to the top-500 firms ofthe Russell 2000 and replace our R2000 indicator with an indicator for thebottom-250 firms of this subsample, as measured using end-of-May marketcapitalization, our IV estimation does not detect an effect of passive ownershipon any of our outcomes. Likewise, we do not find an effect of passive ownershipin a similar placebo test that uses the bottom 500 firms of the Russell 1000.

5.2 Robustness to alternative activism dataAnother commonly used data set in the activism literature is that of Bravet al. (2008) and Brav, Jiang, and Kim (2010). We use SharkWatch as ourprimary data source because it is not limited to hedge fund activists. Thiscan be seen in Table A.9, which lists the activists in each of the two datasets with at least two campaigns; while both data sets cover hedge fundactivists (e.g., Icahn Associates Corp., JANA Partners, Starboard), SharkWatchcovers additional campaigns, including those initiated by pension funds (e.g.,

36



CalPERS, NYC Retirement Systems), individuals (e.g., Karl W. Miller), andnonprofit organizations (e.g., As You Sow). Moreover, additional campaignsfound in SharkWatch represent activist campaigns that do not include a 13Dfiling. Such filings are only required when an activist owns more than 5% of acompany’s equity. While Brav et al. (2008) and Brav, Jiang, and Kim (2010)also make efforts to collect information on campaigns without a 13D filing,their data only include 13 such campaigns during our sample period, whereas185 of the 466 SharkWatch campaigns in our sample lack a 13D filing. Forexample, Sharkwatch covers seven campaigns initiated by Elliott ManagementCo., whereas the Brav et al. (2008) and Brav, Jiang, and Kim (2010) data coveronly five Elliott campaigns. Overall, compared to the 466 campaigns in oursample, only 164 hedge fund campaigns are available during the same periodin the extended data of Brav et al. (2008) and Brav, Jiang, and Kim (2010)15

Our main findings, however, are robust to using activism events definedby Brav et al. (2008) and Brav, Jiang, and Kim (2010). To illustrate this, weanalyze outcomes from this database that are similar to our main results: boardrepresentation is an indicator for if the activist seeks a board seat without aproxy contest; proxy fight is an indicator for the activist attempting to replacethe board through a proxy fight; win/settlement is an indicator for the activisteither winning a proxy fight or achieving a settlement with management;governance objective is an indicator for if the activist seeks governance changes,including the removal of takeover defenses, CEO/chairman replacement, boardindependence; and takeover is an indicator for a takeover bid. Panel A ofTable 11 reports the effects of passive mutual fund ownership on these outcomesbetween 2008 and 2014. Consistent with the results in Tables 5–7, we find thatpassive ownership is associated with an increased likelihood of activists seekingboard representation, both through nonhostile tactics and hostile tactics (i.e.,proxy fights), and an increase in the willingness of managers to settle withactivists. We also find that passive ownership is associated with an increase inthe likelihood of activists seeking governance reform which mirrors the removalof takeover defenses in Table 8.

For takeovers, however, a discrepancy is noted. Using the SharkWatch data,we find higher passive ownership is associated with an increase in acquisitionsby either the activist or a third party (Table 8), but, using the data from Bravet al. (2008) and Brav, Jiang, and Kim (2010), passive ownership and takeoverbids have no relation. This discrepancy seems to originate from a difference inhow the takeover indicator is defined in the two data sets; Brav et al. (2008) andBrav, Jiang, and Kim (2010) measure whether a takeover bid was attempted(Table 11, Column 5), while SharkWatch measures whether a takeover bid was

15 In some cases, however, the Brav et al. (2008) and Brav, Jiang, and Kim (2010) data include campaigns notfound in SharkWatch. These campaigns, however, are often those in which the activist filed a 13D with no statedgoal and took no subsequent actions. SharkWatch includes these type of campaigns from only 50 well-knownactivists (known as the SharkWatch50).

37



Table 11Ownership by passive investors and other activist outcomes using Brav et al. data

Win orBoard settlement Governance Takeover

Dep. variable = representation Proxy outcome objective bid(1) (2) (3) (4) (5)

A. Activism events, 2008–2014Passive % 0.387∗∗ 0.226∗ 0.456∗∗ 0.303∗ −0.171

(0.153) (0.132) (0.182) (0.168) (0.120)


B. Activism events, 1999–2007Passive % −0.072 0.038 −0.243 0.031 0.045

(0.139) (0.117) (0.179) (0.160) (0.048)


This table reports estimates of our instrumental variable estimation used to identify the effect of institutionalownership by passive investors on the tactics of activism campaigns using the extended data of Brav et al. (2008)and Brav, Jiang, and Kim (2010). Specifically, we estimate


n=1

θn[ln(Mktcapit )]n


where Yeit+1 is an indicator for a goal, tactic, or objective of activism event e targeting firm i in year t +1,Passive%it is the percentage of shares outstanding owned by passively managed mutual funds (as defined in theSection 1.1) for stock i at the end of September in year t scaled by its sample standard deviation, Mktcapit isthe CRSP market value of equity of stock i measured at May 31 in year t , and δt are year fixed effects. Theestimation includes an additional control for the natural log of the float-adjusted market value of equity on June30 in year t , ln(Floatit ). The estimate in panel A also includes additional banding controls: an indicator forhaving an end-of-May market capitalization sufficiently close to the cutoff such that the firm will not switchindexes, bandit , an indicator for being in the Russell 2000 last year, R2000it−1, and the interaction of these twoindicators. The campaign outcomes investigated in this table come from Brav et al. (2008) and Brav, Jiang, andKim (2010) and include an indicator for if the activist seeks a board seat without a proxy contest (Column 1); anindicator for whether the activist initiates a proxy fight (Column 2); an indicator for either the activist winningoutright or management settling with the activist rather than accommodating, fighting, or ignoring (Column 3);an indicator for if the activist targets takeover defenses, CEO/chairman replacement, board independence, etc.(Column 4); and an indicator for a takeover bid (Column 5). We instrument Passive% in the above estimationusing R2000it , an indicator equal to 1 if firm i is part of the Russell 2000 index in year t . The sample in panel Aconsists of all activism events that target the top-500 firms in the Russell 2000 index and the bottom-500 firms ofthe Russell 1000 index (i.e., bandwidth = 500) over the 2008-2014 period for which we can obtain holdings datafrom Thomson Reuters Mutual Fund Holdings Database and which we can match with data from the monthlyCRSP file, and the sample in panel B consists of activism events that target the top-500 firms in the Russell indexand the bottom-500 firms of the Russell 1000 index over the 1999-2007 period. The model is estimated usingsecond-order polynomial controls for ln(Mktcap). Standard errors, ε, are clustered at the firm level and reportedin parentheses. *p<.10; **p<.05.

successfully completed (Table 8, Columns 1 and 2). If we rerun our earlieranalysis using takeover outcomes as defined in SharkWatch but restrict oursample to the hedge fund activism events identified in Brav et al. (2008)and Brav, Jiang, and Kim (2010), our findings with respect to takeovers are

38



qualitatively similar to those using the full sample. Our other findings fromTables 5–10 are also robust to restricting our SharkWatch data to the hedgefund activist events found in the Brav et al. (2008) and Brav, Jiang, and Kim(2010).

Unlike the SharkWatch data, which begins coverage in 2006, the data fromBrav et al. (2008) and Brav, Jiang, and Kim (2010), also allows us analyzethe importance of passive ownership for activism in earlier years. Panel B ofTable 11 reports the effect of passive ownership on the same outcomes for1999–2007. We begin this analysis in 1999, because 1998 is the first year inwhich Russell’s float-adjusted market cap is available, and, because this samplepredates the use of banding by Russell, we exclude the additional bandingcontrols. In contrast to the later time period, we do not find an effect of passivemutual fund ownership on any of the outcomes.

The lack of an effect of passive ownership on activists’ strategic choicesin earlier years confirms the growing importance of passive investors forthe governance of firms. The finding is consistent with passive investors’dramatic growth over the last 15 years (see, e.g., Figure 1) yielding themgreater influence over activist campaigns. Moreover, private conversationswith the governance divisions at two of the largest passive institutionssuggest this growing influence is at least partly driven by activists learninghow to win the support of the large passive investors. According to thesepassive institutions, their views on activism have remained relatively constantover time, but that activists have learned through their repeated interactionswith their institutions (that increasingly hold the largest positions in manypublic firms) to tailor their campaign tactics and goals to better reflect thepriorities of long-term investors. Consistent with this, William A. Ackman,a prominent hedge fund manager, recently acknowledged that the growinginfluence of passive investors affects activists’ approaches to campaigns(Ryssdal, Bodnar, and McHenry 2017).16 This shift in tactics might also haveincreased the openness of some institutions to activists’ demands. For example,Dimensional Fund Advisers “rarely engaged with activists before 2007 butformed a corporate governance group that year and started meeting with activistinvestors a few years ago” (Toonkel and Kim 2013).

5.3 Robustness to alternative definitions of passive ownershipFor our analysis above, we measure the ownership stake of passive investorsusing the Thomson Reuters S12 mutual fund data. The S12 data allow for a

16 Ackman stated in an October 2017 interview that “the vast majority of companies in the U.S. today are reallycontrolled by what I would describe as permanent owners of stock. Think index funds like Blackrock. So, theytruly do care about the business over the next 10, 20, 50 years. So the only kind of changes in campaigns we’regoing to run are ones that benefit the business over decades, and those are the only kind of campaigns you canwin. If you have some kind of short-term strategy to make money that’s harmful to the company long-term, you’renot going to get the support of the Blackrocks, the Vanguards and the others” (Ryssdal, Bodnar, and McHenry2017).

39



precise measure of passive ownership. A disadvantage, however, is that the S12mutual fund data do not include the holdings of passive institutional investorslike banks, insurance companies, and pension funds, some of which might alsoadopt passive investment strategies. To analyze whether our use of mutual fundholdings affects our findings, we rerun our analysis using a broader (but noisier)measure of passive ownership, as constructed from the 13F forms reported byinstitutions.

Any financial institution exercising discretionary management of investmentportfolios over $100 million in qualified securities is required to report itsaggregate holdings quarterly to the SEC using Form 13F. The ThomsonReuters Institutional Holdings (13F) Database captures a larger share ofinstitutional ownership than the S12 data; that is, 13F institutional holdingsaccount for about 65% of market capitalization compared to the 26.5% ofmarket capitalization for the S12 mutual fund holdings. Using the 13F data,we classify institutions using Bushee’s (2001) classification. In particular, weclassify “quasi-index” institutions as passive and “transient,” “dedicated,” and“unclassified” institutions as active.

Using the broader measure of passive ownership based on the 13F filingshas no effect on our findings. Using the alternative measure of passiveownership, we repeat our first-stage and IV estimations, and these are reportedin Tables A.10 and A.11. Consistent with the first-stage estimates in Table 3,our first-stage estimates using the 13F data are only significant for the “quasi-index” passive investors (Table A.10, Column 1), and index assignment andmore actively managed institutional holdings have no association (Columns 2–4).17 But, as expected, the increase in passive ownership when using the 13f datais smaller in magnitude (0.9 standard deviations vs. the 1.21 standard deviationsin Table 3, Column 2) and less precisely estimated since “quasi-indexers” alsoincludes the nonpassive holdings of each institution. Despite this limitation, ourIV estimates when using “Quasi-index” remain qualitatively similar to thosereported earlier in Tables 5–10 (see Table A.11).18

6. Conclusion

Recent years have seen a dramatic rise in the amount, the aggressiveness, andthe success rate of activist investing. This increased activism has coincided withthe growing influence of passive institutional investors; as of December 2014,passively managed mutual funds account for more than one-third of all mutualfund assets. In this paper, we ask whether the growing importance of passive

17 The higher “quasi-index” ownership, but no significant relation for other institutional holdings, implies that thehigher passive institutional ownership for stocks in the Russell 2000 corresponds with a lower ownership byshareholders not covered in the institutional level holdings data (e.g., retail investors).

18 In untabulated results, we find that our estimates are also robust to using the definition of passive funds used inBusse and Tong (2012) or the definition of index funds of Iliev and Lowry (2013, 2015).

40



institutional investors has influenced the strategic choices of activists and theirsuccess rates.

Our findings suggest that the growth of passive investors facilitates activism.We find that higher passive ownership is associated with greater success byactivists in obtaining board representation, removing takeover defenses, andfacilitating the sale of a targeted company. We also show that both the likelihoodof activists initiating a proxy fight and winning a settlement are significantlyhigher when passive ownership of the stock is higher. Because such settlementsare typically associated with positive outcomes for shareholders (Bebchuk et al.2017), our findings suggest that higher passive ownership increases activists’ability to pursue policies that increase firm value. Consistent with possibility, wealso find suggestive evidence that announcement returns at the time campaignsare first publicly announced are also more positive than usual when passiveownership is higher. Our findings regarding announcement returns, however,are statistically noisy, possibly reflecting activists’ increased willingness toundertake proxy fights in more marginal cases where where the expected gainsfrom the aggressive engagement tactic are lower. Interestingly, we find littleevidence that higher passive ownership affects the likelihood of an activistcampaign during our sample period; rather, the observed changes in activismthat we find largely occur on the intensive rather than extensive margin. Thisfinding is consistent with the possibility that passive investors might reducethe need for activism among some firms through their low-cost approachesto improving corporate governance (as argued in AGK), while simultaneouslyfacilitating activism among other firms where more costly forms of engagementare necessary.

Our findings appear to be driven by the presence of passive investors ratherthan the increased concentration of firms’ investor base. We find no evidencethat passive ownership is related to activist efforts to affect policies, such aschanges to payout policy or capital structure, which some passive institutionsassociate with shorter-term goals that do not necessarily improve long-termvalue.

Given the myriad of agency conflicts that might exist between managersand shareholders, such as a manager’s inclination to empire build, enjoy thequiet life, or play it safe (e.g., Jensen 1986; Bertrand and Mullainathan 2003;Gormley and Matsa 2016), it is crucial to understand how the shifting nature ofU.S. stock ownership affects the ability of shareholders to discipline managers.Although some worry that the growth of passive investors weakens firm-levelgovernance, our findings provide evidence to the contrary. Specifically, passiveinvestors have been shown to be strong supporters of good governance practicesthat are consistent with long-term firm value (AGK 2016), and we provideevidence here that passive institutional ownership also bolsters the efforts ofactivists that seek similar goals.

41



Figure A.1First-stage and instrumental variable point estimates in the 250 through 750 bandwidths around theRussell 1000/2000 thresholdThis figure plots the point estimates and the 95th percentile confidence intervals by estimation bandwidth choicefor the outcomes reported in Tables 3–10. Appendix Table A.1 gives the variable definitions. The estimationsand the samples are the same as those used in Tables 3–10, except the estimation bandwidth is varied between250 and 750 firms around the Russell 1000/2000 threshold. A second-order polynomial control for ln(Mktcap)is included in all estimations.

42



Table A.1Variable definitions

Variable name Source Definition

R2000 Russell Investments Indicator equal to 1 if firm is in the Russell 2000Mutual fund ownership % Thomson Reuters S12 % of shares outstanding held by mutual funds in

September of year tPassive % Thomson Reuters S12 % of shares outstanding held in September of year t

by passively managed fundsActive % Thomson Reuters S12 % of shares outstanding held in September of year t

by actively managed fundsUnclassified % Thomson Reuters S12 % of shares outstanding held in September of year t

by unclassified fundsSeek board representation SharkWatch (FactSet) Indicator equal to 1 if campaign seeks board

representation for activistMax. val. via policy change SharkWatch (FactSet) Indicator equal to 1 if campaign specifically seeks to

maximize firm valueOther SharkWatch (FactSet) Indicator equal to 1 if campaign does not seek board

rep. or to maximize firm value13D only SharkWatch (FactSet) Indicator equal to 1 if campaign only consists of 13D

filing with no stated goalProxy fight SharkWatch (FactSet) Indicator equal to 1 if campaign includes a proxy fightProxy fight - settlement SharkWatch (FactSet) Indicator equal to 1 if a proxy fight is settledProxy fight - activist wins SharkWatch (FactSet) Indicator equal to 1 if activist wins a proxy fightProxy fight - firm wins SharkWatch (FactSet) Indicator equal to 1 if firm wins a proxy fightProxy fight - withdrawn SharkWatch (FactSet) Indicator equal to 1 if activist withdraws from a

proxy fight before vote occursNumber of seats sought SharkWatch (FactSet) Number of board seats sought by activist in campaign# of seats won in settlement SharkWatch (FactSet) Number of board seats won by activist when a proxy

settlement occursAcquisition by other party SharkWatch (FactSet) Indicator equal to 1 if campaign successfully seeks

and obtains an acquisition by a third partyAcquisition by activist SharkWatch (FactSet) Indicator equal to 1 if campaign successfully seeks an

obtains an acquisition by the activistMerger blocked SharkWatch (FactSet) Indicator equal to 1 if campaign successfully blocks a

merger or agitates for higher priceRemoved takeover defense SharkWatch (FactSet) Indicator equal to 1 if campaign successfully seeks

and obtains removal of takeover defensesIncrease payouts SharkWatch (FactSet) Indicator equal to 1 if campaign successfully seeks

increased payouts to shareholdersCapital structure change SharkWatch (FactSet) Indicator equal to 1 if campaign successfully seeks a

change in capital structureSpinoff and/or divestiture SharkWatch (FactSet) Indicator equal to 1 if campaign successfully seeks a

spinoff or divestitureLawsuit SharkWatch (FactSet) Indicator equal to 1 if campaign features a lawsuitPrecatory proposal SharkWatch (FactSet) Indicator equal to 1 if campaign features a

nonbinding proposalBinding proposal SharkWatch (FactSet) Indicator equal to 1 if campaign features a binding

proposalLetter (nonproxy) SharkWatch (FactSet) Indicator equal to 1 if campaign issues a nonproxy

fight letter to the board or shareholdersSeek reimbursement SharkWatch (FactSet) Indicator equal to 1 if campaign seeks reimbursement

from firmBoard representation Brav et al. data Indicator equal to 1 if campaign seeks board

representation without proxy contestProxy Brav et al. data Indicator equal to 1 if campaign involves a proxy

contestWin or settlement outcome Brav et al. data Indicator equal to 1 if campaign outcome is either a

success or a settlementGovernance objective Brav et al. data Indicator equal to 1 if campaign has a governance

obj. as defined by Brav et al. dataTakeover bid Brav et al. data Indicator equal to 1 if campaign involves a takeover

bid by the activistCAR(-10,10) CRSP Four-factor cumulative abnormal return in 20-day

window around announcement of campaign

Brav et al. data refers to the data of Brav et al. (2008) and Brav, Jiang, and Kim (2010).

43



Table A.2First-stage estimation for ownership by actively managed and unclassified mutual funds

Active % scaled by its Unclassified % scaled by itsDependent variable = sample standard deviation sample standard deviation

(1) (2) (3) (4) (5) (6)

R2000 −0.127 0.191 0.206 0.168 0.081 0.191(0.377) (0.390) (0.401) (0.294) (0.340) (0.327)

Polynomial order, N 1 2 3 1 2 3Banding controls Yes Yes Yes Yes Yes YesFloat control Yes Yes Yes Yes Yes YesYear fixed effects Yes Yes Yes Yes Yes YesObservations 466 466 466 466 466 466R-squared .113 .138 .138 .123 .125 .131

This table reports estimates of our first-stage regression of ownership by actively managed and unclassifiedmutual funds onto an indicator for membership in the Russell 2000 index plus additional controls over the 2008-2014 sample period. The specification is the same as that used in Table 3, except that the dependent variablein Columns 1-3 is now Active%it , which is the percentage of shares outstanding owned by actively managedmutual funds for stock i at the end of September in year t scaled by its sample standard deviation, and thedependent variable in Columns 4-6 is now Unclassified%it , which is the percentage of shares outstanding ownedby unclassified mutual funds for stock i at the end of September in year t scaled by its sample standard deviation.Both Active% and Unclassified% are defined in Section 1.1. Standard errors are clustered at the firm level andreported in parentheses.

Table A.3Ownership by passive investors and analyst coverage

Dependent variable = Number of analysts covering stock

(1) (2) (3)

Passive % −0.847 −1.202 −1.132(1.983) (1.798) (1.830)

Polynomial order, N 1 2 3Banding controls Yes Yes YesFloat control Yes Yes YesYear fixed effects Yes Yes YesObservations 466 466 466

This table reports estimates of our instrumental variable estimation used to identify the effect of institutionalownership by passive investors on the number of analysts covering a stock. Specifically, we estimate

Yeit+1 =α+β Passive%it +N!

n=1

θn[ln(Mktcapit )]n


where Yeit+1 is the number of analysts covering firm i at the time an activism event targeting that firm in yeart +1, Passive%it is the percentage of shares outstanding owned by passively managed mutual funds (as definedin Section 1.1) for stock i at the end of September in year t scaled by its sample standard deviation, Mktcapitis the CRSP market value of equity of stock i measured at May 31 in year t , and δt are year fixed effects. Theestimation includes an additional control for the natural log of the float-adjusted market value of equity on June30 in year t , ln(Floatit ). The estimate also includes additional banding controls: an indicator for having an end-of-May market capitalization sufficiently close to the cutoff such that the firm will not switch indexes, bandit ,an indicator for being in the Russell 2000 last year, R2000it−1, and the interaction of these two indicators. Weinstrument Passive% in the above estimation using R2000it , an indicator equal to 1 if firm i is part of the Russell2000 index in year t . The sample consists of the top-500 firms of the Russell 2000 index and the bottom-500 firmsof the Russell 1000 over the 2008–2014 period for which we obtain holdings data from Thomson Reuters MutualFund Holdings Database and which we match with data from the monthly CRSP file. The model is estimatedusing polynomial order controls N =1, 2, and 3 for ln(Mktcap). Standard errors, ε, are clustered at the firm leveland reported in parentheses.

44



Table A.4Robustness of findings to selecting sample only using end-of-May market cap rankings

Board-related tactics and outcomes Other activism outcomes Performance

Seek Acquired Acquired Removedboard Proxy Seats Proxy [by third [by Merger takeover CAR

Dependent variable = rep. fight sought settlement party] activist] blocked defense (−10,10)(1) (2) (3) (4) (5) (6) (7) (8) (9)

Passive % 0.371∗∗ 0.290∗∗ 1.016∗ 0.191∗ 0.101 0.029 −0.264∗∗ 0.092 0.112(0.185) (0.145) (0.544) (0.105) (0.070) (0.048) (0.124) (0.060) (0.091)


This table reports estimates of the second-stage regression of our instrumental variable estimation to identifythe effect of institutional ownership by passive investors on our activism outcome variables when we select oursample using firms with an end-of-May market cap ranking between 500 and 1500. The estimation and outcomesare the same as those used in Tables 5-10. We instrument Passive% using R2000it , an indicator equal to 1 if firmi is part of the Russell 2000 index in year t . The model is estimated using activism campaigns over the 2008-2014period that target firms in the selected sample and includes a second-order polynomial control for ln(Mktcap).Standard errors, ε, are clustered at the firm level and reported in parentheses. *p<.10; **p<.05.

Table A.5Robustness of findings to including firm-year observations without activism events

Board-related tactics and outcomes Other activism outcomes

Seek Acquired Acquired Removedboard Proxy Seats Proxy [by third [by Merger takeover

Dependent variable = rep. fight sought settlement party] activist] blocked defense(1) (2) (3) (4) (5) (6) (7) (8)

Passive % 0.022∗ 0.023∗∗ 0.065∗ 0.015∗∗ 0.009∗ 0.004∗ −0.014 0.003(0.012) (0.010) (0.036) (0.007) (0.005) (0.002) (0.011) (0.002)

Polynomial order, N 2 2 2 2 2 2 2 2Banding controls Yes Yes Yes Yes Yes Yes Yes YesFloat control Yes Yes Yes Yes Yes Yes Yes YesYear fixed effects Yes Yes Yes Yes Yes Yes Yes YesObservations 6,803 6,803 6,803 6,803 6,803 6,803 6,803 6,803

This table reports estimates of the second-stage regression of our instrumental variable estimation to identify theeffect of institutional ownership by passive investors on our activism outcome variables when do not limit oursample to firm-years where an activism event occurs. The estimation and outcomes are the same as those usedin Tables 5-9, and outcomes for firm-year observations with no activism event are coded as 0. For example, afirm that is not targeted by an activist that year is also not subject to an activist-driven proxy fight. We instrumentPassive% using R2000it , an indicator equal to 1 if firm i is part of the Russell 2000 index in year t . The modelis estimated using all firm-year observations over the 2008-2014 period in the selected sample and includes asecond-order polynomial control for ln(Mktcap). Standard errors, ε, are clustered at the firm level and reportedin parentheses. *p<.10; **p<.05.

45



Table A.6Robustness of findings to controlling for liquidity



Dependent variable = rep. fight sought settlement party] activist] blocked defense (-10,10)(1) (2) (3) (4) (5) (6) (7) (8) (9)

Passive % 0.369∗∗ 0.353∗∗ 1.024∗∗ 0.231∗∗ 0.128∗ 0.059∗ −0.152∗∗ 0.046 0.125(0.160) (0.143) (0.519) (0.104) (0.078) (0.033) (0.070) (0.039) (0.080)


This table reports estimates of the second-stage regression of our instrumental variable estimation to identify theeffect of institutional ownership by passive investors on our activism outcome variables when we add controls forliquidity. The estimation and outcomes are the same as those used in Tables 5-10, except we include two additionalcontrols for liquidity, the Amihud measure of illiquidity and the bid-ask spread. We instrument Passive% usingR2000it , an indicator equal to 1 if firm i is part of the Russell 2000 index in year t . The model is estimatedusing activism campaigns that target firms in the selected sample over the 2008-2014 period and includes asecond-order polynomial control for ln(Mktcap). Standard errors, ε, are clustered at the firm level and reportedin parentheses. *p<.10; **p<.05.

Table A.7Robustness of findings to excluding activist campaigns that include a 13D filing only




Passive % 0.473∗∗ 0.415∗∗ 1.231∗∗ 0.260∗∗ 0.139 0.075∗ −0.218∗∗ 0.077 0.122(0.195) (0.169) (0.595) (0.120) (0.087) (0.043) (0.107) (0.050) (0.088)


This table reports estimates of the second-stage regression of our instrumental variable estimation to identify theeffect of institutional ownership by passive investors on our activism outcome variables when we exclude activistcampaigns that only include a 13D filing. The estimation and outcomes are the same as those used in Tables 5-10.We instrument Passive% using R2000it , an indicator equal to 1 if firm i is part of the Russell 2000 index in yeart. The model is estimated using activism campaigns that target firms in the selected sample over the 2008-2014,excluding those with only a 13D filing, and includes a second-order polynomial control for ln(Mktcap). Standarderrors, ε, are clustered at the firm level and reported in parentheses. *p<.10; **p<.05.

46



Table A.8Robustness of findings to clustering standard errors at the activist level




Passive % 0.355∗∗ 0.321∗∗ 0.941∗ 0.200∗∗ 0.110 0.060∗ −0.174∗∗ 0.057∗ 0.115∗(0.147) (0.127) (0.483) (0.088) (0.069) (0.034) (0.072) (0.033) (0.070)


This table reports estimates of the second-stage regression of our instrumental variable estimation to identify theeffect of institutional ownership by passive investors on our activism outcome variables when we instead clusterour standard errors at the activist level. The estimation and outcomes are otherwise the same as those used inTables 5-10. We instrument Passive% using R2000it , an indicator equal to 1 if firm i is part of the Russell 2000index in year t . The model is estimated using activism campaigns over the 2008-2014 period that target firmsin the selected sample and includes a second-order polynomial control for ln(Mktcap). Standard errors, ε, areclustered at the activist level and reported in parentheses. *p<.10; **p<.05.

47



Table A.9Activists and number of campaigns by data source

SharkWatch # Brav et al. (2008) and Brav, Jiang, and Kim (2010) #

GAMCO Asset Management, Inc. 70 GAMCO INVESTORS, INC. ET AL 16CtW Investment Group 18 ICAHN CARL C 11Starboard Value LP 17 JANA PARTNERS LLC 8Icahn Associates Corp. 16 RELATIONAL INVESTORS LLC 8The California Public Employees

Retirement System15 Starboard Value LP 7

New York City Retirement Systems 13 VA PARTNERS I, LLC 6ValueAct Capital Management LP 12 VA Partners I, LLC 6JANA Partners LLC 8 ELLIOTT ASSOCIATES, L.P. 5Relational Investors, LLC 8 ATLANTIC INVESTMENT

MANAGEMENT INC5

Clinton Group, Inc. 8 BLUE HARBOUR GROUP, LP 4Elliott Management Corporation 7 Blue Harbour Group, L.P. 4Calvert Investment Management, Inc. 5 Corvex Management LP 3Biglari Capital Corp. 5 MHR FUND MANAGEMENT LLC 3Sandell Asset Management Corp. 5 SPO ADVISORY CORP 3P. Schoenfeld Asset Management LP

(New York)4 PAULSON & CO INC 3

Corvex Management LP 4 Mount Kellett Capital Management LP 3Shamrock Partners Activist Value Fund

LLC4 HARBINGER CAPITAL PARTNERS

MASTER FUND I, LTD.2

Land & Buildings InvestmentManagement LLC

4 THIRD POINT LLC 2

Miller/Howard Investments, Inc. 4 Luxor Capital Group, LP 2Engaged Capital LLC 4 INTEGRATED CORE STRATEGIES

(US) LLC2

Praesidium Investment Management Co.LLC

3 BLUE HARBOUR GROUP, L.P. 2

MCM Management, LLC 3 CLINTON GROUP INC 2Blue Harbour Group LP 3 FAIRHOLME CAPITAL

MANAGEMENT LLC2

Sarissa Capital Management LP 3 MMI INVESTMENTS, L.P. 2Harbinger Capital Partners 3 Engaged Capital LLC 2International Brotherhood of Teamsters 3 PERSHING SQUARE CAPITAL

MANAGEMENT, L.P.2

Barington Companies Investors LLC 3 RAMIUS LLC 2Seneca Capital Investments LP 3 SHAMROCK ACTIVIST VALUE FUND

L P2

As You Sow 3 THIRD AVENUE MANAGEMENT LLC 2Pershing Square Capital Management LP 3 Blue Harbour Group, LP 2Marcato Capital Management LLC 3 BREEDEN CAPITAL MANAGEMENT

LLC2

UNITE HERE 3 MARCATO CAPITAL MANAGEMENTLP

2

Cal. State Teachers Retirement SystemRelational Investors, LLC

3

Davis Selected Advisers LP 3TPG-Axon Management LP 2PepsiCo, Inc. 2Corvex Management LP Related Fund

Management LLC2

Steel Partners, L.L.C. 2Breeden Capital Management LLC 2Third Point LLC 2Eminence Capital LLC 2Karl W. Miller 2Perry Corp. (New York) 2Charles Robert Palmer 2Southeastern Asset Management, Inc. 2TRT Holdings, Inc. 2

(Continued)

48



Table A.9(Continued)

SharkWatch # Brav et al. (2008) and Brav, Jiang, and Kim (2010) #

Prospect Capital Corporation 2MHR Fund Management LLC 2Atlantic Investment Management, Inc. 2CR Intrinsic Investors LLC SAC Capital

Advisors, LLC2

Investment Partners Asset Management,Inc.

2

Crescendo Advisors LLC 2Carlson Capital LP 2Millennium Management LLC 2MSMB Capital Management LLC 2Luxor Capital Group LP 2SAC Capital Advisors, LLC 2Nierenberg Investment Management

Company, Inc.2

Continental Grain Company 2Fairholme Capital Management LLC 2Green Century Capital Management, Inc. 2Neuberger Berman LLC 2Validus Holdings, Ltd. 2Pentwater Capital Management LP 2Orange Capital LLC 2

This table reports the activists and number of campaigns found in the two different activism databases,SharkWatch, Brav et al. (2008), and Brav, Jiang, and Kim (2010). For brevity, the list for each activism data set islimited to activists with at least two campaigns in our main estimation sample. The sample consists of all activismevents that target the top-500 firms in the Russell 2000 index and the bottom-500 firms of the Russell 1000 index(i.e., bandwidth = 500) over the 2008-2014 period for which we can obtain holdings data from Thomson ReutersMutual Fund Holdings Database and which we can match with data from the monthly CRSP file.

49



Table A.10Impact of index assignment on types of 13F institutional ownership

Dependent variable = Ownership % scaled by its sample standard deviation:

Quasi-indexers Transient Dedicated Unclassified(1) (2) (3) (4)

R2000 0.903∗∗ −0.068 −0.347 −0.293(0.355) (0.403) (0.342) (0.310)

Polynomial order, N 2 2 2 2Banding controls Yes Yes Yes YesFloat control Yes Yes Yes YesYear fixed effects Yes Yes Yes YesObservations 466 466 466 466R-squared .222 .190 .045 .195

This table reports estimates of a regression of instiutional ownership as reported in 13F filings on an indicatorfor membership in the Russell 2000 index plus additional controls. Specifically, we estimate

Ownership%it =η+λR2000it +N!

n=1

χn[ln(Mktcapit )]n


where R2000it is a dummy variable equal to 1 if stock i is in the Russell 2000 Index at end of June in year t ,Mktcapit is the CRSP market value of equity of stock i measured at May 31 in year t,N is the polynomial orderwe use to control for ln(Mktcapit ), and δt are year fixed effects. The estimation includes an additional controlfor the natural log of the float-adjusted market value of equity on June 30 in year t , ln(Floatit ). The estimate alsoincludes additional banding controls: an indicator for having an end-of-May market capitalization sufficientlyclose to the cutoff such that the firm will not switch indexes, bandit , an indicator for being in the Russell 2000 lastyear, R2000it−1, and the interaction of these two indicators. Ownership%it measures institutional ownership (inpercent) for stock i at the end of September in year t . In this table we use four different definitions for Ownership%for stock i: (1) the percentage of shares outstanding owned by “quasi-indexer” institutions; (2) the percentageof shares outstanding owned by “transient” institutions; (3) the percentage of shares outstanding owned by“dedicated” institutions; and (4) the percentage of shares outstanding owned by “unclassified” institutions. Theinstitution classifications are defined in Bushee (2001). The sample consists of all activism events that target thetop-500 firms in the Russell 2000 index and the bottom-500 firms of the Russell 1000 index (i.e., bandwidth= 500) over the 2008-2014 period for which we obtain holdings data from Thomson Reuters Mutual FundHoldings Database and which we match with data from the monthly CRSP file. The model is estimated using apolynomial order control for ln(Mktcap) of N =2. Standard errors, u, are clustered at the firm level and reportedin parentheses. **p<.05.

50



Table A.11Robustness of findings to using 13F data and quasi-index ownership



Dependent variable = rep. fight sought settlement party] activist] blocked defense (−10,10)(1) (2) (3) (4) (5) (6) (7) (8) (9)

Quasi-index % 0.474∗ 0.429∗∗ 1.256∗ 0.266∗ 0.146 0.080∗ −0.232∗ 0.077 0.138(0.248) (0.214) (0.713) (0.141) (0.092) (0.045) (0.138) (0.053) (0.094)


This table reports estimates of the second-stage regression of our instrumental variable estimation to identifythe effect of ownership by institutions classified as “quasi-indexers” by Bushee (2001) on our activism outcomevariables. The estimation and outcomes are the same as those used in Tables 5-10. We instrument Quasi-index%using R2000it , an indicator equal to 1 if firm i is part of the Russell 2000 index in year t . The model is estimatedusing activism campaigns that target firms in the selected sample over the 2008-2014 and includes a second-orderpolynomial control for ln(Mktcap). Standard errors, ε, are clustered at the firm level and reported in parentheses.*p<0.10; **p<.05.

51



References

Ackman, W. A. 2016. William A. Ackman to Shareholders of Pershing Square CapitalManagement, LP, January 26, 2016. Pershing Square Capital Management, LP, New York.https://assets.pershingsquareholdings.com/2014/09/Pershing-Square-2015-Annual-Letter-PSH-January-26-2016.pdf.

Activist Insight. 2014. Activist Insight Monthly (September).

Angrist, J., and J. Pischke. 2009. Mostly harmless econometrics: An empiricist’s companion. Princeton, NJ:Princeton University Press.

Aquila, F. J. 2015. Negotiating a settlement with an activist investor. Practical Law The Journal, April.https://www.sullcrom.com/files/upload/Apr15_InTheBoardroom.pdf

Appel, I. R., T. A. Gormley, and D. B. Keim. 2016. Passive investors, not passive owners. Journal of FinancialEconomics 121:111–41.

———. 2015. Identification using Russell 1000/2000 index assignments: A discussion of methodologies.Working Paper.

Bakke, T., and T. Whited. 2012. Threshold events and identification: A study of cash shortfalls. Journal of Finance67:1083–111.

Bebchuk, L. 2007. The myth of the shareholder franchise. Virginia Law Review 93:675–732.

Bebchuk, L., A. Brav, and W. Jiang. 2015. The long-term effects of hedge fund activism. Columbia Law Review115:1085–156.

Bebchuk, L., A. Brav, W. Jiang, and T. Keusch. 2017. Dancing with activists. Working Paper.

Becht, M., J. Franks, C. Mayer, and S. Rossi. 2009. Returns to shareholder activism: Evidence from a clinicalstudy of the Hermes UK Focus Fund. Review of Financial Studies 22:3093–129.

Benoit, D. 2014a. Blackrock’s Larry Fink: Typical activists are too short-term. Wall Street Journal, January 16.https://blogs.wsj.com/moneybeat/2014/01/16/blackrocks-larry-fink-typical-activists-are-too-short-term/

Benoit, D. 2014b. CEOs test: Contending with activist investors. Wall Street Journal, November 25.https://www.wsj.com/articles/ceos-test-contending-with-activist-investors-1416957944

Benoit, D., and K. Grind. 2015. Activists investors’ secret ally: Big mutual funds. Wall Street Journal, August9. https://www.wsj.com/articles/activist-investors-secret-ally-big-mutual-funds-1439173910

Bertrand, M., and S. Mullainathan. 2003. Enjoying the Quiet Life? corporate governance and managerialpreferences. Journal of Political Economy 111:1043–1075.

Biers, J. 2014. Sotheby’s gives hedge fund activist Dan Loeb board seats, avoiding shutdown. Artdaily.org,May 6. http://artdaily.com/news/69908/Sotheby-s-gives-hedge-fund-activist-Dan-Loeb-board-seats–avoiding-showdown#.W5bB0JMzowt.

Bird, A., and S. A. Karolyi. 2016. Do institutional investors demand public disclosure? Review of FinancialStudies 29:3245–77.

Boone, A., and J. White. 2015. The effect of institutional ownership on firm transparency and informationproduction. Journal of Financial Economics 117:508–33.

Boyson, N., N. Gantchev, and A. Shivdasani. 2017. Activism mergers. Journal of Financial Economics 126:54–73.

Boyson, N., and N. Pichler. 2018. Hostile resistance to hedge fund activism. Review of Financial Studies. AdvanceAccess published May 17, 2018, 10.1093/rfs/hhy058.

Bradley, M., A. Brav, I. Goldstein, and W. Jiang. 2010. Activist arbitrage: A study of open-ending attempts ofclosed-end funds. Journal of Financial Economics 95:1–19.

Brav, A., A. Dasgupta, and R. Mathews. 2015. Wolf pack activism. Working Paper.

52



Brav, A., W. Jiang, and H. Kim. 2010. Hedge fund activism: A review. Foundations and Trends in Finance4:1–66.

———. 2015a. Recent advances in research on hedge fund activism: Value creation and identification. AnnualReview of Financial Economics 7:579–95.

———. 2015b. The real effects of hedge fund activism: Productivity, asset allocation, and labor outcomes.Review of Financial Studies 28:2723–69.

Brav, A., W. Jiang, S. Ma, and X. Tian. 2018. How Does Hedge Fund Activism Reshape Corporate Innovation?Journal of Financial Economics. Advance Access published May 17, 2018, 10.1093/rfs/hhy058.

Brav, A., W. Jiang, F. Partnoy, and R. Thomas. 2008. Hedge fund activism, corporate governance, and firmperformance. Journal of Finance 63:1729–75.

Bushee, B. 2001. Do institutional investors prefer near-term earnings over long-run value? ContemporaryAccounting Research 18:207–46.

Busse, J., and Q. Tong. 2012. Mutual fund industry selection and persistence. Review of Asset Pricing Studies2:245–74.

Briggs, T. 2007. Corporate governance and the new hedge fund activism: An empirical analysis. Journal ofCorporation Law 32:681–738.

Chang, Y., H. Hong, and I. Liskovich. 2015. Regression discontinuity and the price effects of stock marketindexing. Review of Financial Studies 28:212–46.

Clifford, C. 2008. Value creation or destruction? Hedge funds as shareholder activists. Journal of CorporateFinance 14:323–36.

Coffee, J., and D. Palia. 2016. The wolf at the door: The impact of hedge fund activism on corporate governance.Annals of Corporate Governance 1:1–94.

Collin-Dufresne, P., and V. Fos. 2015. Do prices reveal the presence of informed trading? Journal of Finance70:1555–82.

Crane, A., S. Michenaud, and J. Weston. 2016. The effect of institutional ownership on payout policy: Evidencefrom index thresholds. Review of Financial Studies 29:1377–408.

Denes, M., J. Karpoff, and V. McWilliams. 2015. Thirty years of shareholder activism: A survey of empiricalresearch. Journal of Corporate Finance 44:405–24.

Dimson, E., O. Karakas, and X. Li. 2015. Active ownership. Review of Financial Studies 28:3225–68.

Fich, E., J. Harford, and A. Tran. 2015. Motivated monitors: The importance of institutional investors’ portfolioweights. Journal of Financial Economics 118:21–48.

Fos, V. 2017. The disciplinary effects of proxy contests. Management Science 63:655–71.

Fos, V., and M. Tsoutsoura. 2014. Shareholder democracy in play: Career consequences of proxy contests.Journal of Financial Economics 114:316–40.

Gantchev, N. 2013. The costs of shareholder activism: Evidence from a sequential decision model. Journal ofFinancial Economics 107:610–31.

Gillan, S., and L. Starks. 2007. The evolution of shareholder activism in the United States. Journal of AppliedCorporate Finance 19:55–73.

Gompers, P., J. Ishii, and A. Metrick. 2010. Extreme governance: an analysis of dual-class firms in the UnitedStates. Review of Financial Studies 23:1051–88.

Gormley, T., and D. Matsa. 2016. Playing it safe? Managerial preferences, risk, and agency conflicts. Journal ofFinancial Economics 122:431–55.

Greenwood, R., and M. Schor. 2009. Investor activism and takeovers. Journal of Financial Economics 92:362–75.

53



Grossman, S. J., and O. Hart. 1980. Takeover bids, the free-rider problem, and the theory of the corporation. BellJournal of Economics 11:42–64.

Guo, R., T. Kruse, and T. Nohel. 2008. Undoing the powerful anti-takeover force of staggered boards. Journalof Corporate Finance 14:274–88.

Henderson, M. T., and D. Shapiro Lund. 2017. Index funds are great for investors, risky for corporate governance.The Wall Street Journal, June 22. https://www.wsj.com/articles/index-funds-are-great-for-investors-risky-for-corporate-governance-1498170623

Hoffman, L., and D. Benoit. 2015. Activist investors ramp up, and boardroom rifts ensue. The Wall Street Journal,April 16. https://www.wsj.com/articles/activists-ramp-up-and-boardroom-rifts-ensue-1429209792

Iliev, P., and M. Lowry. 2013. Are mutual funds active voters? Working Paper.

———. 2015. Are mutual funds active voters? Review of Financial Studies 28:446–85.

Investment Technology Group. 2008. 2008 Russell reconstitution recap. Report.

Jensen, M. 1986. Agency costs of free cash flow, corporate finance and takeovers. American Economic Review76:323–29.

Kahan, M., and E. B. Rock. 2015. Hedge funds in corporate governance and corporate control. In Institutionalinvestor activism: Hedge funds and private equity, economics and regulation. Eds. W. W. Bratton andJ. A. McCahery. Oxford, U.K.: Oxford University Press.

Kedia, S., L. Starks, and X. Wang. 2017. Institutional investors and hedge fund activism. Working Paper.

Klein, A., and E. Zur. 2009. Entrepreneurial shareholder activism: Hedge funds and other private investors.Journal of Finance 64:187–229.

Kumar, R., and R. O’Hanley. 2016. Protecting the interests of long-term shareholders in activistengagements. Harvard Law School Forum on Corporate Governance and Financial Regulation, October17. https://corpgov.law.harvard.edu/2016/10/17/protecting-the-interests-of-long-term-shareholders-in-activist-engagements/

Lipton, M. 2017. Dealing with activist hedge funds and other activist investors. HarvardLaw School Forum on Corporate Governance and Financial Regulation, January 26.https://corpgov.law.harvard.edu/2017/01/26/dealing-with-activist-hedge-funds-and-other-activist-investors/

McCahery, J. A., Z. Sautner, and L. T. Starks. 2016. Behind the scenes: The corporate governance preferencesof institutional investors. Journal of Finance 71:2905–32.

Sodali, M. 2017. Institutional Investor Survey 2017. https://www.morrowsodali.com/eventassets/1488278692-Morrow%20Sodal_%20Institutional%20investor%20survey%202017.pdf

McRitchie, J. 2012. Use of exempt solicitations up dramatically in 2012: Chesapeake (CHK) latest example.CorpGov.net, May 30. http://www.corpgov.net/2012/05/use-of-exempt-solicitations-up-dramatically-in-2012-chesapeake-chk-latest-example/

Mullins, W. 2014. The governance impact of index funds: Evidence from a regression discontinuity. WorkingPaper.

Popadak, J. 2014. A corporate culture channel: How increased shareholder governance reduces firm value.Working Paper.

Russell Investments. 2013. Russell U.S. equity indexes construction and methodology, June 2013. Report.https://www.ftse.com/products/indices/russell-us

Ryssdal, K., B. Bodnar, and S. McHenry. 2017. Why Bill Ackman sees activists investing as a moralcrusade. Marketplace, October 31. https://www.marketplace.org/2017/10/31/business/why-bill-ackman-sees-activist-investing-moral-crusade

Schmidt, C., and R. Fahlenbrach. 2017. Do exogenous changes in passive institutional ownership affect corporategovernance and firm value? Journal of Financial Economics 124:285–306.

54



Sommer, J. 2013. Challenging management (but not the market). New York Times, March 16.

Stock, J., J. Wright, and M. Yogo. 2002. A survey of weak instruments and weak identification in generalizedmethod of moments, Journal of Business & Economics Statistics 20:518–29.

The Economist. 2015. Capitalism’s unlikely heroes: Why activist investors are good for the public company. TheEconomist, February 7. https://www.economist.com/leaders/2015/02/05/capitalisms-unlikely-heroes

Toonkel, J., and S. Kim. 2013. Activist investors find allies in mutual, pension funds. Reuters,April 9. https://www.reuters.com/article/us-funds-activist/activist-investors-find-allies-in-mutual-pension-funds-idUSBRE9380DU20130409

Wei, W., and A. Young. 2017. Selection bias or treatment effect? A re-examination of Russell 1000/2000 Indexreconstitution. Working Paper.

Wilcox, J. 2005. Shareholder nominations of corporate directors: Unintended consequences and the case forreform of the U.S. proxy system. In Shareholder access to the corporate ballot. Ed. L. Bebchuk. Cambridge,MA: Harvard University Press.

55

standing on the shoulders of giants: the effect of passive ... · [14:32 19/10/2018...

Documents